AI-Assisted Editorial Work

The report has been developed through the author’s many years of work on municipal governance, sustainable and inclusive cities, institutional analysis, moral biology, and regenerative practice, as well as through iterative dialogue with generative AI systems.

AI has supported articulation, structure, comparison, summarisation, and revision. Lars A. Engberg has selected, assessed, and edited the material and bears responsibility for the report’s claims, interpretations, omissions, and proposals.

An AI system can influence, structure, and in practice direct human actions. Legal, professional, moral, and democratic responsibility cannot, however, be transferred to the system. It must continue to be borne by identifiable people and legitimate institutions.

Field Temperature

A work field can be operating in green, yellow, or red.

Green operations do not mean an absence of busyness or conflict. They mean that the task can be carried with professional judgement, relational continuity, capacity for correction, and regeneration intact.

Yellow operations mean that the field is still functioning, but increasingly through hidden compensation, rework, dependence on key individuals, or growing capacity debt.

Red operations are not a judgement of the employee. They are a signal that the system may no longer be able to carry the core task, rights, or the function of care without harm or persistent human overload.

Red is a demand that the system respond.

Green, yellow, and red must not be used as HR, pay, absence, disciplinary, or performance data. The raw registration may constitute personal data. Field temperature must therefore be designed with short retention periods, minimum group sizes, aggregation, no individual histories, and an explicit prohibition on reuse for personnel assessment.

The person must be protected. The field must be legible. The system must be able to respond.

AI as a Bounded Operational Actor

The Sophia Lumen way of working begins with human judgement, relationship, and responsibility. AI can support articulation, search, comparison, drafting, scenarios, and — under stronger conditions — reversible process steps.

Report 01 distinguishes between five levels of mandate:

1. reflection and analysis;
2. drafting and recommendation;
3. reversible execution under fixed criteria;
4. bounded process orchestration;
5. highly sensitive or irreversible decisions, in which legitimate authority and the final decisive impulse remain human and institutional.

The greater the mandate, the greater the risk of automation bias, responsibility fog, and a moral deformation zone in which the nearest employee is blamed for a system they do not genuinely control. Operational delegation must therefore never exceed the organisation’s ability to understand, control, stop, and repair.

Aligned Intelligence is used here in a deliberately embodied and institutional sense. The fixed point is not the machine, its future trajectory, or an abstract intelligence. The fixed point is the living: the body, the nervous system, the relationship, the place, and the people who may be affected by a decision. AI must align itself with this primacy — not the other way round.

The order of priority is therefore:

1. dignity and fundamental rights;
2. the democratically decided core task;
3. the citizen’s actual life and opportunity for correction;
4. professional judgement and responsibility;
5. the field’s carrying capacity;
6. the public economy as a whole;
7. the long-term good of the commons;
8. operational efficiency;
9. possible budget savings.

The Phoenix Economy

The public economy must not be read as one large automatable cost base. Denmark’s total public expenditure was provisionally DKK 1,472.8 billion in 2025. Social protection, health, education, and public services are to a large extent the very substance of the welfare state.

AI’s relevant economic power lies in reducing unnecessary friction, connecting knowledge, detecting errors earlier, protecting human capacity, and preventing more costly breakdowns.

In 2024, Danish municipalities had approximately 418,300 full-time employees and a total payroll of DKK 220 billion. As pure scale scenarios, protecting or releasing:

• 0.10 per cent of capacity corresponds to approximately 418 full-time equivalents and a payroll equivalent of DKK 220 million;
• 0.25 per cent corresponds to approximately 1,046 full-time equivalents and DKK 550 million;
• 0.50 per cent corresponds to approximately 2,092 full-time equivalents and DKK 1.1 billion.

These are not forecasts or cash savings. Danish cases nevertheless show that substantial local gains can be documented: Ringkøbing-Skjern has calculated 37,400 released hours over three years; Vejle’s screen-visit pilot calculated 68.1 hours per citizen annually and a total of approximately 2,800 hours, or 2.4 full-time equivalents, for about 40 citizens; and Copenhagen’s user-management case shows why a large gross effect can become a much smaller net gain after development, operations, and continuing manual work.

The report distinguishes between:

• G1 — released or protected capacity;
• G2 — documented avoided cost or harm;
• G3 — shared strategic and democratic capacity to act.

The gain must be corrected for implementation, training, control, maintenance, error correction, new work, and burdens displaced onto others. Only then can the municipality decide:

1. what should repair capacity debt;
2. what should protect carrying capacity;
3. what can be shared with other pressured fields or with prevention;
4. and what may potentially become budgetary room for action.

Democratic Horizon

When field temperature, citizen experience, and economic effects can move through the system without making people transparent, the municipality can become better at sensing, learning, and correcting democratically.

In this movement, the citizen is not only a user, client, or source of data. Knowing From the Ground points towards the citizen as observer, bearer of knowledge, corrective, co-interpreter, and a gradually more sovereign democratic participant. Data should be offered, not extracted. AI can help people understand, formulate, and follow public processes, but must not become the superior interpreter of their experience.

The welfare state is not reborn by delivering more with ever less living capacity. It is reborn by releasing capacity from unnecessary friction and returning it to the core task, relationships, rights, and shared capacity to act.

Report 01 does not offer a finished future welfare state. It shows how this force can begin to work under the right forms and in right relationship within one concrete municipal work field.

1.1 Cross-Pressure as an Operational State

Municipal operations work under simultaneous demands that cannot be reduced to a single cause: legislation and reform tracks; documentation and control; recruitment, absence, and staff turnover; fiscal frameworks and budget culture; complex citizen pathways; digital systems; climate, nature, and long-term resilience.

Cross-pressure is not in itself a sign of poor organisation. It is a fundamental condition. The problem arises when the organisation loses the ability to distinguish between the complexity genuinely inherent in the task and the friction produced by the design of work itself.

The task rarely disappears when the framework tightens. The child must still be met. The citizen must still receive a decision. The older person must still receive help. The road and sewer must still function. The water must still be managed. The cultural centre must still be able to hold a shared space. When formal capacity is reduced, the burden is displaced onto employees’ extra work, citizens’ waiting time, relatives’ coordination, lower relational continuity, more defensive documentation, or deferred maintenance and nature expenditure.

It is therefore insufficient to ask only whether the organisation complies with budget and deadlines. A field can deliver many services while consuming its own future carrying capacity.

1.2 Green, Yellow, and Red Operations

Green, yellow, and red are field indicators, not HR categories.

In green operations, the field can carry its task. Responsibility is sufficiently clear. Professional judgement can be exercised. Errors are discovered and corrected. Documentation has a comprehensible function. Load is followed by an opportunity for regulation and recovery. Green operations do not mean an absence of conflict or busyness; they mean functioning with capacity for correction and regeneration intact.

In yellow operations, the field functions increasingly through compensation. Rework grows. Informal solutions become permanent. Dependence on particular key people increases. Handover becomes weaker. Documentation becomes more defensive. There is still life in the field, but its resilience is declining.

In red operations, formal functioning is maintained through persistent overload, loss, or risk. People become permanent buffers. Errors cannot be properly investigated. Stop options exist only on paper. The citizen, a relative, or a single employee becomes the actual coordinator. Documentation protects the system more than the right. A red state can still produce green numbers on selected indicators. This is precisely why averages and productivity figures must never stand alone.

1.3 Field Temperature — Protecting the Person, Making the Field Legible

Field temperature is a collective assessment of the conditions of the work field. It is not a diagnosis of the individual employee and not an indirect measure of mood, loyalty, resilience, or productivity.

An employee provides a situated signal: Can this field currently carry the task responsibly? The signal may concern time, professional discretion, relational continuity, transfer of responsibility, the possibility of correcting errors, capacity debt, or risk of harm.

The legal and organisational boundary must be formulated precisely. A raw response may constitute personal data if it can be connected to the employee who provided it. Pseudonymisation is not the same as anonymisation. It is therefore insufficient simply to declare that the colours “are not HR data”.

The correct design requirement is:

The purpose of field temperature is not to assess the individual employee, and the system must be designed technically, organisationally, and legally so that individual signals cannot be reused as HR, control, absence, or performance data.

A responsible practice requires, among other things:

• no management access to individual responses;
• no permanent individual temperature histories;
• no linkage with appraisal interviews, pay, absence, productivity, or personnel cases;
• no mandatory free text that can identify people;
• minimum group sizes and suppression of small groups;
• short retention of raw signals;
• aggregated patterns rather than individual profiles;
• shared governance with the working-environment organisation and employee representation;
• clear information about purpose, access, retention, and use;
• the right not to respond when a response cannot be given truthfully or safely.

Field temperature must not be inferred through emotion recognition, biometric analysis, tone of voice, keystrokes, facial expression, sentiment analysis of emails, or hidden behavioural profiling. It must be a consciously provided assessment of working conditions — not machine reading of a person’s inner life.

The individual signal becomes organisational knowledge only through aggregation, interpretation, and feedback. A colour without shared inquiry can easily become just another management figure. A field must therefore be able to describe:

• what makes it green, yellow, or red;
• which conditions lie within local capacity to act;
• which conditions cut across departments;
• which conditions follow from political priorities, legislation, or financing;
• what action the signal triggers;
• and what happened afterwards.

An organisation in which employees risk sanction or suspicion for reporting red does not have field temperature. It has a green-reporting mechanism.

Management must reward truthful sensing, not comfortable signals.

1.4 From the Field’s Signal to Democratic Responsibility

Field temperature changes the responsibility of management.

Management is not about returning the indicator to green. Management is about investigating what the signal says about the core task and taking responsibility for the conditions that lie within the organisation’s power.

A red pattern may indicate a need for more resources. But it may also reveal:

• conflicting rules;
• a broken workflow;
• weak coordination;
• an unsuitable digital system;
• an unclear mandate;
• discontinued maintenance;
• a task that no longer creates public value;
• or a central-government reform whose actual implementation burden has been underestimated.

Red should therefore trigger a fixed but proportionate process:

1. The field describes the load without identifying individual employees.
2. Management investigates whether the cause is local, cross-cutting, political, or external.
3. Economics, law, professional practice, citizen experience, and the working environment are read together.
4. A relief measure, correction, or reversible intervention is selected.
5. Responsibility, timeframe, and feedback are made visible.
6. The temperature is followed after the intervention.
7. If the cause lies in central-government regulation or financing, the signal is passed onwards.

The municipal council should not see individual responses. It should be able to see institutional patterns: which core tasks are under persistent pressure, which priorities create conflicting demands, where savings merely displace burden, and where the organisation uses people as a buffer between statutory expectations and inadequate conditions.

Field temperature is therefore more than management information.

It is democratic institutional sensing.

2.1 Rhythm

All municipal work fields have rhythms: days, weeks, school years, budget years, holiday periods, seasons, political cycles, acute events, and longer pathways. Home care does not have the same rhythm as a planning department. A child’s development does not follow the authority’s case calendar. A local community can take years to build and disappear within a few months.

When work is organised as though load were constant, a permanent state of exception can easily arise. Rhythm may therefore require protected overlaps, coherent professional time, periods for preparation, seasonally adjusted capacity, and the possibility of digesting difficult conversations or intensive periods.

Rhythm is not a soft addition to operations. It is an operational condition.

2.2 Load and Regulation

Load is not in itself harmful. A living system can handle intensity if it has a clear function, beginning, and end, and if the field can return to a sustainable state.

Many small, unfinished loads can be more erosive than one intense period: constant interruptions; unclear transfers of responsibility; repeated errors; systems requiring permanent manual compensation; moral conflicts without a legitimate route; tasks that never feel complete.

Regulation is therefore more than a break. It is the field’s possibility of re-establishing coherence, attention, and capacity to act after load.

2.3 Nourishment

A work field is nourished by what makes good task performance possible: professional trust, a clear mandate, usable tools, collegial continuity, comprehensible decisions, the possibility of learning from errors, access to relevant knowledge, time for an important conversation, and management capable of receiving bad news.

Nourishment is not the same as comfort. A field can be full of conflict and still be nourishing if conflicts can be spoken openly and lead to understanding or correction. An apparently harmonious field can be depleted if problems have become unsayable.

2.4 Boundaries and Diversity

The municipality consists of boundaries between authority and provider, professions, departments, the state, region, municipality, private suppliers, digital systems, and the citizen’s life. Some boundaries protect rights, confidentiality, or responsibility. Others create loss because relevant knowledge and responsibility cannot pass through.

A healthy boundary is permeable to what must move and protective of what must not be extracted.

Municipal practice requires legal knowledge, professional experience, the citizen’s own experience, technical understanding, economic overview, relational sensing, and knowledge of place. None of these alone can carry complex work fields. A field becomes fragile when one form of knowledge gains a monopoly on defining reality.

2.5 Regeneration

Regeneration does not mean that everything must grow. It means that the field can rebuild the capacity it uses.

This may require removing a registration requirement, abolishing a meeting, rolling back an automation, ending a temporary solution, or sharing a responsibility. A field does not necessarily regenerate by receiving yet another project. It may regenerate by being allowed to stop compensating for something that no longer serves the task.

3.1 The Work of Translation

Municipal work takes place between different languages: those of law, economics, professional practice, technology, politics, and the citizen’s concrete experience. The employee translates a rule into an action, a system output into a comprehensible answer, a citizen’s story into an administrative category, and professional doubt into something to which the organisation can respond.

The translation is not mechanical. It requires understanding both the system and the life in which the system must work. When time is reduced or standardisation becomes too tight, the words may arrive while the meaning is lost.

3.2 The Citizen, Relatives, and Civil Society as Buffer

The citizen becomes a buffer when they must find the responsible person, repeat information, understand contradictory messages, discover errors, or coordinate between systems. Friction hits particularly hard when the person is simultaneously ill, under financial pressure, in crisis, or without strong linguistic, digital, or social resources.

Relatives and civil society carry significant parts of the social infrastructure. They help with translation, continuity, observation, participation, and relational repair. But community must not be reduced to free reserve capacity. When the municipal task functions only because others take over coordination and care without sufficient support or influence, responsibility has been displaced in a hidden way.

3.3 Rotation Before Heroes

Many work fields are carried by key people who know whom to contact, how the systems actually work, and how a difficult case or local relationship can be moved forward. Their contribution should be recognised. But an organisation that functions only through heroes is not robust.

Rotation does not mean that everyone is interchangeable. It means that no one should be forced permanently to carry the organisation’s missing connections. Knowledge and responsibility must be able to move through overlap, partnership, shared reflection, and gradual handover. Relational continuity should be preserved, while structural dependence on one person should be reduced.

Rotation before heroes is both care and governance.

4.1 Flow

Flow is the necessary movement through the work field: information, understanding, responsibility, professional assessment, consent, physical assistance, financial support, relational continuity, or knowledge of the next step.

Flow is not the same as speed. A slow pathway can have good flow if time is used for necessary observation, consultation, or professional investigation. A fast pathway can have poor flow if meaning, context, or responsibility is lost.

Not everything should flow. Confidential information must be protected. A child’s story should not circulate merely because more people are interested. An employee’s personal reflection should not automatically become shared data. Good flow makes the necessary mobile and keeps the sensitive protected.

4.2 Friction

Friction is not every form of resistance. Consultation of a party, professional disagreement, safety control, and time for relationship can protect something legitimate.

Friction is the work, waiting time, or resistance that binds capacity without protecting a right, quality, or safety and without moving the task correspondingly forward. It may be repetition with no new function, waiting without responsibility, rework after known errors, handovers without preserved context, documentation nobody uses, or systems requiring permanent manual repair.

Friction is also a trace. A private list may reveal that the shared system does not carry the necessary memory. An overloaded key person may reveal that the connections have become person-borne. A growing document may reveal that responsibility and risk are unclear.

4.3 Sensitivity

Sensitivity concerns how much can be lost, who bears the loss, how easily it is detected, and how difficult it is to repair.

A factual error may perhaps be corrected quickly. An error affecting a child’s trust, a citizen’s housing, an older person’s medication, or an irreversible condition in nature can have much greater and longer consequences.

An average gain is insufficient if a smaller group is exposed to severe harm. Sensitivity makes the distribution visible.

4.4 The Triad in Practice

A work field can begin with six questions:

1. What must be able to move for the task to succeed?
2. What meaning must not be lost?
3. Where are time, attention, or relational capacity bound?
4. What legitimate function does the resistance serve?
5. What can be lost, and who bears the consequence?
6. Where must a human opportunity to stop and correct be possible?

The triad does not close the inquiry. It opens it.

5.1 Moral Strain

Moral strain arises when an employee repeatedly experiences a gap between what is judged professionally or humanly right and what is actually possible. It may mean ending a conversation too early, sending a letter expected to create uncertainty, watching a pathway fragment without a mandate to bring it together, or correcting the same systematic error case by case.

Moral strain can lead to distance, mechanical language, silence, cynicism, or excessive personal assumption of responsibility. Both extremes are signs of a field that no longer distributes responsibility and load responsibly.

5.2 Decency as an Operational State

Decency is operational. It consists in providing a comprehensible reason, avoiding unnecessary humiliation, stating who is responsible, protecting confidential matters, saying what the municipality does not yet know, and making correction possible.

The municipality can say no and still act decently. It can impose requirements and still protect dignity. Decency is not the opposite of authority, but a quality of legitimate exercise of public authority.

5.3 Documentation as Protection

Documentation can protect rights, professional assessments, continuity, reviewability, and organisational memory. It has a living function when it helps the field remember, understand, coordinate, justify, correct, and learn.

The report distinguishes between necessary documentation, rights-protecting documentation, learning documentation, and defensive production of control.

5.4 Documentation as Armour

Documentation becomes armour when text and registration primarily absorb organisational uncertainty or displace responsibility. The same information is registered in several places. Standard texts grow to cover every conceivable situation. More approvals are introduced because no one has a mandate to decide. The concrete assessment disappears in the volume of text.

The paradox of armour is that more documentation can make responsibility less visible.

The citizen must see through a larger mass of text. The employee must produce and check more. The organisation may still lack the precise documentation that would actually protect the right or enable learning.

When generative AI makes text cheaper to produce, this risk grows. A rapid draft is not a gain if it creates longer documents, more versions, and a greater need for control.

The companion report The Regenerative Municipality gathers the digital and administrative control test in the Orwell Test:

Does the solution create control without corresponding insight, counter-power, and the possibility of repair?

The test does not replace the concrete legal, professional, or democratic assessment. It makes visible at an early stage when more registration, automation, or documentation increases institutional control without increasing the affected person’s understanding, capacity to act, or access to correction.

Three stop questions remain relevant:

Does this become more text or more clarity?

Are we moving responsibility away from an identifiable person?

Are we making the field more repairable or more armoured?

6.1 Service, Capital, and Transfers

The municipality’s economic categories have different relationships to the work field.

Service and operations contain a large share of human and relational capacity: employee time, teaching, care, exercise of public authority, administration, culture, and daily maintenance. When the service-expenditure ceiling tightens without the task correspondingly disappearing, people easily become the buffer.

Capital expenditure creates buildings, roads, systems, and physical infrastructure. Capital investment can strengthen future operations, but it can also impose persistent costs or friction on operations. A solution that is inexpensive to build may be costly in maintenance, control, or relational repair work.

Transfers are not ordinary municipal production. They affect people’s material ability to live, eat, travel, and participate. An administrative delay can therefore rapidly become social and health-related escalation.

The categories must not be mixed when gains are calculated. Released employee time in service provision is not automatically a cash saving. A capital investment is not an operational gain until its total effect is understood. A change in transfers cannot be assessed solely as municipal administrative economics.

6.2 The Service-Expenditure Ceiling as a Field Boundary

The service-expenditure ceiling functions as a real boundary on how much municipal operation can be financed. It may be necessary as a form of macroeconomic governance, but its effects become concrete locally.

When the task, statutory requirements, and citizen needs remain while the framework tightens, efficiency does not arise automatically. Intensification, hidden overtime, lower relational quality, more standardisation, and fewer opportunities for prevention and correction may arise instead.

The economic question is therefore not only whether expenditure has remained within the ceiling. It is also what the ceiling makes the field do with people’s time, professional practice, and relationships.

6.3 Demographics, Equalisation, Reimbursement, and Liquidity

Municipalities’ room for action is affected by population composition, social and geographical structure, the equalisation system, reimbursement rules, and liquidity. In the municipal introduction landscapes, these mechanisms are addressed through dated official sources; their concrete effect must always be assessed in the local economic and organisational context.

Demographics change the composition and rhythm of tasks. Equalisation and reimbursement affect where costs and incentives are placed. Liquidity affects the municipality’s ability to invest before breakdown and carry transition costs. A municipality may have a long-term opportunity for gain while lacking the short-term financial capacity to realise it.

A technical business case is therefore never the entire basis for a decision.

6.4 Economics as Public Prioritisation

Economics is not only a constraint. It is also a language of prioritisation. A budget determines whether there is overlap between shifts, time for professional reflection, continuity around children and older people, maintenance before breakdown, early local involvement, or the possibility of repairing an error.

Scarcer resources do not make work ecology less relevant. They make it more important to distinguish between necessary public capacity and friction that merely binds resources.

6.5 The Scale of the Public Economy

Denmark’s total public expenditure was provisionally DKK 1,472.8 billion in 2025. The distribution by purpose shows that social protection accounted for 41 per cent, health for 17 per cent, education for 12 per cent, and general public services for 12 per cent.

Converted into approximate amounts:

Source: Statistics Denmark, General Government Accounts 2022–2025, published 3 June 2026. The amounts are the report’s rounded calculations based on the published shares.

The figures must not be used to portray the entire public sector as one large automatable cost. A large share of the amount is the very substance of the welfare state: income security, care, treatment, teaching, rights, defence, emergency preparedness, and infrastructure.

The relevant question is therefore not:

What percentage of DKK 1,472.8 billion can AI save?

The relevant question is:

Where in the total economy is capacity bound in unnecessary friction, where is expensive repair of earlier failure produced, and where can released capacity be returned to direct welfare and democratic/regenerative capacity to act?

Even small shares illustrate the scale of the system. Mathematically, 0.1 per cent of total public expenditure corresponds to approximately DKK 1.47 billion; 0.5 per cent to DKK 7.36 billion; and 1 per cent to DKK 14.73 billion. These amounts are not measures of effect or savings forecasts. They show only why small systemic changes can become significant — and why documentation discipline must be strict.

6.6 Five Flows in the Public Budget

Official public accounts are divided by authority, function, and account. The Phoenix account adds a cross-cutting reading of what the expenditure does.

A. Direct welfare

Care, income security, teaching, treatment, protection, mobility, access, and rights. This flow must not be made efficient out of existence.

B. Necessary institutional infrastructure

Professional documentation, legal safeguards, supervision, democratic decision-making, data security, redundancy, maintenance, and control of power. It can be improved, but not simply removed.

C. Unnecessary friction

Double registration, manual transfers, repeated enquiries, incompatible systems, unclear responsibilities, defensive documentation, and waiting time without public value.

D. The repair economy

Reprocessing, complaints, acute interventions, temporary staffing, legal proceedings, breakdown, escalation of conflict, and rebuilding lost relationship or trust.

E. Regenerative and democratic capacity

Prevention, local observation, shared learning, citizen participation, open methods, public technological sovereignty, care for places and commons, and the institution’s ability to correct itself.

In economic terms, the Phoenix movement consists in shifting resources from unnecessary friction and avoidable repair towards direct welfare and regenerative/democratic capacity — without hollowing out necessary institutional infrastructure.

6.7 The Municipal Labour Market as Living Capacity

In 2024, Danish municipalities had approximately 528,200 employees, corresponding to 418,300 full-time employees, and a total payroll of DKK 220 billion. The average payroll equivalent is therefore approximately DKK 526,000 per full-time equivalent. The figure is an anchor of scale, not a complete marginal cost.

Source: Local Government Denmark (KL), The Municipal Labour Market in Figures 2024. The scenarios are calculations made by this report and must not be read as expected AI effects.

The 2026 agreement on municipal finances sets a service-expenditure ceiling of DKK 336.8 billion, total municipal expenditure of DKK 501.8 billion in the agreement’s balance, and an administrative reduction of DKK 250 million. The lower end of the February report’s illustrative scenario — an average capacity equivalent of five full-time equivalents across 98 municipalities — corresponds to approximately 490 full-time equivalents and a payroll equivalent of around DKK 258 million.

The similarity in order of magnitude is not evidence that technology can or should deliver the administrative reduction in the financial agreement. It shows the difference between two logics:

• An external budget requirement determines the reduction first.
• The Phoenix account first documents whether net capacity actually exists, who produced it, what the field needs, and which part can be allocated without harm.

6.8 Sickness Absence as a Sign of Capacity Debt — Not as an Automatic Source of Savings

Average municipal sickness absence was 13.9 working days per full-time employee in 2024. In elder care, health, and disability services, it was 16.3 working days; in administration, 9.1.

Across 418,300 full-time employees, 13.9 days correspond mathematically to approximately 5.8 million days of absence. A change of one average day would correspond to approximately 418,000 working days or around 1,900 full-time equivalents of capacity at 220 working days.

This is not a savings potential, and the report claims no causal relationship between AI, field temperature, and national sickness absence. Employees continue to receive pay, causes are many, and part of any possible effect would appear as continuity, lower pressure from temporary staffing and overtime, and reduced load on colleagues.

The figures instead show the magnitude of the capacity dimension that the condition of the work field may have, and why field temperature is of interest as a possible early indicator before capacity debt becomes visible in absence, turnover, errors, and breakdown.

7.1 Friction as a Hidden Budget Item

Friction becomes economics when it binds hours, creates reprocessing, produces meeting inflation, increases absence, leads to complaints or escalation, and makes the organisation dependent on costly acute repair.

It may be visible in:

• repeated citizen contacts;
• reopened cases;
• document versions;
• waiting time between functions;
• overtime and absence;
• staff turnover;
• complaints and remittals;
• parallel lists and systems;
• time spent by relatives and citizens;
• deferred maintenance;
• conflicts arising because participation came too late.

Friction must not be confused with all time-consuming work. Time for relationship, consultation, protection of rights, professional disagreement, maintenance, and observation can be productive public capacity. The question is what function the time serves.

Friction is not always unintended. The report distinguishes between:

• protective friction, which safeguards rights, professional thoroughness, and democratic counter-power;
• unintended friction, arising through system breaks, repetition, and unclear responsibility;
• distributive friction, which deliberately or in practice restricts access or moves work and risk onto particular people.

The final form cannot be treated solely as a process problem. It requires political visibility: Who decided that the burden should exist, what purpose does it serve, and who bears its price?

7.2 The Per-Citizen Layer and Scale

Per-citizen figures can make municipalities comparable, but they can also conceal differences in demographics, geography, task profile, social structure, and organisational design.

The per-citizen layer should therefore be used for orientation, not as a ranking. It must be accompanied by a description of what the amount contains, which tasks are particularly demanding, and which costs lie outside the municipality’s own accounts.

Scale matters. Five minutes of unnecessary work in a rare pathway is not the same as five minutes in a high-volume field. But scale alone cannot determine priority; a rare error may be more sensitive and irreversible.

7.3 The Follow-the-Money Card

A light-touch card can contain:

The card should be capable of completion using existing knowledge and a small number of supplementary observations. If the mapping requires a new and extensive registration machine, it contradicts its own purpose.

7.4 Six Simultaneous Accounts

A work field cannot be assessed through a single economic number. The Phoenix account holds six accounts side by side.

The accounts must not be compressed into one monetary total. Some entries are kroner, others hours, events, distributive effects, or non-compensatory rights requirements.

A positive budget entry cannot compensate for systematic deterioration of fundamental rights or for a work field being operated persistently in red.

8.1 A Living Initial State

A baseline is a sufficient description of the work field before the intervention. It may include numbers of repeated contacts, waiting time, handovers, document versions, control time, errors, and citizen understanding. But it must also examine how the task is performed: hidden overtime, dependence on key people, relational continuity, and the possibility of correction.

A small, honest baseline is better than a large, falsely precise measurement.

8.2 The Intervention as a Hypothesis

An intervention must be formulated as a hypothesis:

If we change X in this bounded work field, we expect Y to improve without Z deteriorating.

Z is decisive. A faster process is not a gain if legal safeguards, comprehensibility, relational quality, or nature’s long-term carrying capacity are weakened.

The intervention must have an identifiable responsibility-holder, a clear scope, stop criteria, and a practical alternative or previous work path.

8.3 Net Correction

Net correction asks:

• What improved?
• What deteriorated?
• What did not change?
• What capacity did the intervention require?
• Where did the work or risk move?
• Who received the gain, and who bore the cost?
• Which simultaneous conditions affected the result?
• What can the material not tell us?

A digital solution can reduce internal processing time and create more work for citizens. A shorter text may require more oral explanation. An optimised route may weaken continuity. Net correction follows the entire relevant field.

8.4 G1 — Released or Protected Capacity

G1 is less rework, fewer repetitions, better handovers, more coherent professional time, better relational continuity, or less dependence on hidden compensation.

Released time is not automatically a budget saving. It may be used for existing backlogs, quality, prevention, relationship, training, or recovery.

8.5 G2 — Avoided Cost or Harm

G2 may consist of fewer errors, complaints, escalations, reopened pathways, breakdowns, acute interventions, or irreversible losses in nature and place.

It must be clear whether the gain is observed, rendered plausible, calculated, or solely a design hypothesis. A hypothetical future harm must not be presented as a realised saving.

8.6 G3 — Strategic Room for Action and Robustness

G3 arises when the municipality gains a genuine possibility of choosing differently: investing in prevention, maintenance, relational continuity, professional depth, local commons, or ecological capacity.

G3 does not exist merely because a spreadsheet shows a possibility. A legitimate decision-maker must actually be able to reallocate the capacity.

8.7 Regenerative Reinvestment — the Gain Must Have a Route Back

A municipality is not a nervous system. Municipal administrations do, however, perform nervous-system functions.

They register changes in people, places, and tasks. They carry signals forward, connect experience with law, economics, and professional practice, prioritise, act, discover effects, correct, and learn. A healthy work field must not only be able to react quickly. It must also be able to inhibit, distinguish, recover, and repair.

AI can support this capacity. It can reduce repetition, make information easier to find, support formulation, and release time from particular workflows. But faster signal processing is not the same as better regulation.

If all released capacity is immediately converted into more tasks, higher production requirements, or a general budget reduction, AI can make a yellow or red field work faster without making it greener. The gain may be real while its use is extractive.

The following rule therefore applies:

No budget gain before field gain.

Before an AI or process gain is treated as economic room for action, the municipality must be able to show:

1. that the gain remains after implementation, training, control, error correction, and new or displaced work;
2. that quality, relationship, legal safeguards, professional judgement, and the possibility of correction have not deteriorated;
3. that the field is sufficiently stable for the capacity actually to be allocated;
4. that the use of the gain has been decided visibly and legitimately.

The regenerative order is:

8.8 No Double Counting and Non-Compensatory Conditions

The same effect must not be registered as both released time, avoided payroll expenditure, and new budgetary room for action without separate documentation.

Nor may certain conditions be compensated away in one combined calculation. A time gain cannot automatically outweigh weakened legal safeguards, lack of human control, hidden profiling, loss of confidentiality, or irreversible loss in nature.

An intervention can be green on one indicator and red on another. The red state must not be dissolved in the average.

8.9 A Shared 90-Day Pilot Format

Where a fixed pilot period is useful, Report 01 uses the same 90-day rhythm as its companion report The Regenerative Municipality. Report 01 retains its operational step logic but gathers it into four decision gates:

Some work fields require shorter or longer observation. Any deviation must then be justified by the rhythm and sensitivity of the task. The 90 days are a shared municipal pilot format, not a requirement that biological, relational, or physical effects must be fully visible by day 90.

8.10 Monday Start — the Smallest Thing That Can Be Done Without Lying

A municipality does not need to wait for a large project in order to begin. A legitimate first step can be undertaken as a short, shared field reading:

1. Choose one type of task. Delimit the beginning, the end, and the desired living outcome.
2. Read Flow · Friction · Sensitivity for 30 minutes. What must move, where is capacity bound, and what may be significantly lost?
3. Choose a few baseline proxies from existing data. For example, repeated contacts, handovers, reprocessing, waiting time, or a simple green/yellow/red field assessment.
4. Formulate one reversible intervention. State what is expected to improve and what must not deteriorate.
5. Agree the stop, responsibility, and next decision. Who can respond, where is the alternative route, and when will the pathway be assessed again?

The Monday start is not a miniature implementation. It is a test of whether the field can become legible without creating a new burden of measurement or meetings. It may conclude that the municipality is not ready for a pilot because red operations, an unclear mandate, or inadequate correction capacity must first be addressed. That is also a legitimate outcome.

8.11 Six Recurring Worked Examples

The worked examples show the full movement from baseline to net correction, field test, and reinvestment decision.

Figures marked illustrative scenario are assumptions made by this report. Figures marked empirical anchor come from a published source. No case may be transferred directly to another municipality or workflow.

8.12 The Practice Toolbox

Appendices A–H make the report’s way of working directly usable. The templates may be copied, shortened, and adapted. They share three rules:

• existing data and professional knowledge come before new registration;
• all assumptions, calculations, and uncertainties must be visible;
• a field is completed only when it helps to understand, protect, correct, or decide.

9.1 What Must Be Transformed

• documentation that primarily protects the system from responsibility;
• budget boundaries that reward local suboptimisation;
• digitalisation that shifts work onto citizens and relatives;
• control loops without learning;
• technology that accelerates red operations;
• supplier dependence that privatises public knowledge;
• gain requirements that harvest capacity before the field is stable;
• reforms without a functioning return path from implementation.

9.2 What Must Be Carried Through the Transformation

• universal rights;
• human dignity;
• equality and solidaristic risk sharing;
• professional judgement;
• democratic legitimacy;
• local rootedness;
• institutional memory;
• public responsibility;
• the human right not to be made transparent;
• shared ownership of learning, data, and fundamental methods.

9.3 The Regenerative Movement

An improvement does not become regenerative merely by increasing productivity.

It must move capacity:

• from rework to doing it right the first time and comprehensibility;
• from late repair to prevention;
• from isolated control to shared learning;
• from hidden human compensation to visible organisational responsibility;
• from supplier lock-in to public and shareable capacity;
• from opaque harvesting to democratic decisions about gains;
• from intensification to relationship, professional practice, and shared capacity to act.

9.4 The Non-Compensatory Boundary

A positive savings figure cannot outweigh:

• systematic weakening of fundamental rights;
• absence of a human route for correction;
• discriminatory effects;
• hidden surveillance;
• persistent red operations;
• or a practice in which the citizen’s access to help depends on surrendering more of their life than the task requires.

The Phoenix Economy is therefore not a maximisation model. It is a public balance with boundaries.

9.5 The Reform Test

A public reform or AI intervention can be called regenerative when it can answer five questions:

1. Core task: Which concrete welfare or rights effect improves?
2. The field: Does the work become more sustainable, correctable, and less dependent on hidden heroic work?
3. The total economy: Are all significant new costs and displacements visible?
4. Democracy: Can affected people understand, contest, and influence the movement?
5. The route of the gain: Where does the released capacity go, and who decides?

The Phoenix is therefore not a story about returning to an earlier welfare state. It is a story about preserving the moral core and allowing the institutional form to re-emerge with stronger sensing, better feedback, more truthful accounts, and intelligence aligned with the best interests of the commons.

The Field’s Living Rhythm

The field follows the transitions of the day, the movements of the school year, and the child’s developmental windows. It is nourished by familiar adults, belonging, play, learning, time for observation, and coherence between home, school, and other environments. Regeneration does not mean merely that a case is closed, but that the child and the adults again have the possibility of learning, regulating themselves, and entering sustainable relationships. An early sign of red operations is that everyone reacts to the latest event while no one can any longer hold the child’s pathway as a whole; the child or family then becomes the bearer of continuity.

Economic Cycle

The field’s economy includes more than case-processing time and interventions. Repeated assessments, parallel plans, long waiting times, school absence, family strain, and later escalation can move costs between school, social services, health, family, and future budgets.

An early investment in capacity is not automatically a documented G2 gain. But the follow-the-money card must make visible when a short-term saving creates more extensive work later.

Flow · Friction · Sensitivity

Flow consists of the child’s and family’s experience, professional observations, responsibility, support, and understanding of the next step. Friction may consist of repeated assessments, parallel action plans, unclear ownership, and long periods without visible action. Sensitivity is high because errors can affect safety, development, learning, family relationships, and trust.

Illustrative Interventions

A bounded pathway may test one clear responsibility-holder, a handover with preserved context, or a double communication layer in which the legal text is accompanied by a short explanation borne by a responsible human. AI can support structure, plain language, and comparison, but must not replace the child’s voice, relational observation, or concrete professional assessment.

The inner life of the child and family is not raw material. There must be strong restraint regarding emotion profiling, recordings, secondary use of conversations, and linkage of data across purposes.

Economic Illustration

In a highly sensitive field, an intervention may release only a few hours and still have great public value. Example C in section 8.11 shows how a modest net G1 can be accompanied by earlier correction, better understanding, and a possibly avoided escalation without exaggerating these effects into a certain savings figure.

The Field’s Living Rhythm

The field follows meals, sleep, medication, mobility, courses of illness, and the effects of the seasons on the body. It is nourished by recognisable relationships, professional overlap, the possibility of detecting small changes, and planning able to bend around a concrete person. Regeneration may mean that the citizen regains function, security, or everyday rhythm; it may also mean that a stable and dignified level can be maintained. The field moves towards red operations when the plan consistently takes precedence over the body and every deviation is treated as an operational disturbance rather than professional information.

Economic Cycle

The field’s economy includes staffing, routes, documentation, aids, prevention, rehabilitation, and interfaces with the region, doctor, and relatives. A planning gain may create G1 in one place while increasing hospital contact, errors, or relatives’ work elsewhere.

Continuity may have economic value through earlier detection and less rework, but it must not be reduced to a guaranteed saving. It is also a quality of the service itself.

Flow · Friction · Sensitivity

Flow consists of relevant observations, changes in need, responsibility, medical and care-related information, and the citizen’s preferences. Friction consists of changes, inadequate handovers, double registration, telephone chains, and timetables with no room for variation. Sensitivity includes medication, falls, nutrition, loneliness, safety, and dignity.

Illustrative Interventions

A field may test a continuity window, a clear handover route for non-acute changes, or a flexible marking of visits that required professionally justified additional time. AI can support transcription, structure, and planning proposals, but no journal entry or professional prioritisation may become uncontrolled automatic action.

The citizen’s home and body must not be treated as freely available data environments. More sensor data is not automatically more care. It must be clear who responds and what the citizen can genuinely understand and contest.

Economic Illustration

Example B in section 8.11 shows a field in which a documented time gain is deliberately reinvested in overlap, follow-up, and familiar relationships. A cash saving does not necessarily arise, but the work field gains greater capacity to detect changes and act professionally.

The Field’s Living Rhythm

The field contains simultaneous temporalities. A payment deadline may be acute, while healing, work identity, and the way back to participation are rarely linear. It is nourished by comprehensible language, predictable contact, material stability, and room to correct a misunderstood picture. Regeneration means that the person’s agency increases without concealing legal requirements or authority responsibility. A sign of red operations is that the control chain accelerates while the citizen’s actual possibility of understanding, responding, and acting diminishes.

Economic Cycle

Transfers, reimbursement, employment measures, social support, and housing are connected, but follow different economic and legal logics. An administrative saving can create greater social harm or later municipal expenditure. Conversely, a clear and rapid factual correction can prevent escalation without being a cash saving in the traditional sense.

The field’s economy must therefore follow municipal time, the citizen’s material situation, and costs shifted to other sectors or future periods.

Flow · Friction · Sensitivity

Flow consists of comprehensible information, relevant documentation, the citizen’s own explanation, financial assistance, responsibility, and requests for correction. Friction consists of repeated documentation requirements, unclear letters, many contact people, parallel plans, and standard pathways that do not fit the situation. Sensitivity is high because errors can affect income, housing, health, and family.

Illustrative Interventions

A concrete field may test a first letter as an action map, a rapid correction route for bounded factual errors, or one shared formulation of the next step. AI can support plain language, disposition, and comparison, but must not classify people as motivated or unwilling, create hidden risk profiling, or stand as an authority before the citizen.

A person must not be required to surrender more of their life than the concrete public task requires.

Economic Illustration

A highly sensitive authority field should not be forced into the same productivity measure as a high-volume field. Example C in section 8.11 shows how comprehensibility, factual correction, and lower risk of escalation may be more important than a large reduction in hours.

The Field’s Living Rhythm

The field follows both political and administrative deadlines and longer material rhythms: rain, groundwater, growth, decomposition, maintenance, and the slow change of places. It is nourished by site visits, early operational knowledge, real alternatives, and the possibility of allowing new information to change direction. Regeneration may consist of restored connection, reduced vulnerability, or maintenance that prevents later breakdown. The field moves towards red operations when irreversible decisions are locked before significant knowledge has found a legitimate route in.

Economic Cycle

The field’s economy connects capital and operations. An inexpensive capital solution may create costly maintenance, climate risk, or loss of nature. Deferred maintenance can improve the short-term account and worsen the long-term one. Irreversible losses cannot be reduced to a single budget item.

Net correction must therefore follow the entire lifetime and distribution of risk: who receives the gain, who receives the operational task, and which costs are shifted to future budgets, citizens, or nature.

Flow · Friction · Sensitivity

Flow consists of technical and ecological knowledge, local experience, alternatives, responsibility, and physical flows such as water and mobility. Friction consists of systems and maps that cannot be understood together, professional contributions arriving too late, unclear transitions between project and operations, and participation processes without real decision space. Sensitivity includes safety, water, nature, housing, heat, noise, local identity, and irreversible conditions.

Illustrative Interventions

A pathway may test a living place map, an early correction window, or an operational review before the capital decision is locked. AI can support document search, thematic analysis of consultation responses, and scenario visualisation, but cannot decide the meaning of a place, represent what is missing from the dataset, or transform a model into democratic legitimacy.

The Field’s Living Rhythm

The field has a high-frequency rhythm with daily waves of contact, statutory deadlines, payment days, seasonal peaks, and sudden events. It is nourished by precise shared knowledge, clear transfers of responsibility, the possibility of human contact, and administrative time to complete an action correctly. Regeneration means that fewer enquiries are caused by the system’s own lack of clarity and that the front line can again function as a sensing entrance rather than a permanent repair workshop. Red operations appear when the first point of contact must absorb contradictions for which no underlying function takes responsibility.

Economic Cycle

The field is often a high-volume domain. Small frictions can therefore accumulate significantly. Repeated contacts, re-entry, and unclear status bind municipal time and citizen time. But efficiency must not be achieved merely by shifting work to people with less capacity to bear it.

Reduced contact time is not a gain if the citizen subsequently contacts another unit or loses access to a right.

Flow · Friction · Sensitivity

Flow consists of the question, identification, relevant information, responsibility, documents, payment, and status. Friction consists of unclear digital entrances, telephone chains, standard replies, repeated data entry, and a system status that does not correspond to the real process. Sensitivity varies; an apparently small administrative error can affect deadlines, benefits, identity, or access to help.

Illustrative Interventions

A field may test one truthful next step, a repetition loop that investigates the cause of renewed contact, or a short administrative friction week directed at the workflow. AI can support search, plain language, and preliminary categorisation, but must not pretend to be a responsible human, rank citizens by tone, or close the human route in sensitive situations.

Economic Illustration

In a high-volume field, a few minutes of reduced rework can accumulate into a significant gross release. Example A in section 8.11 shows why implementation and control must first be deducted and why only part of net G1 necessarily becomes allocable budgetary room for action.

The Field’s Living Rhythm

The field follows seasons, local traditions, rehearsal pathways, volunteers’ lives, and the slow building of trust. It is nourished by accessible places, continuity, artistic and local autonomy, and the possibility of maintaining what already works. Regeneration may consist of generational succession, new connections, and a community able to share responsibility without making particular people indispensable. The field moves towards red operations when project cycles, documentation, and informal care work require more capacity than the practice itself is allowed to rebuild.

Economic Cycle

The domain’s budgets may be small in relation to the relational infrastructure they carry. Short project periods, extensive application requirements, and constant demands for innovation can force actors to use capacity for financing and documentation rather than practice.

Volunteers and artists may become informal social workers or local coordinators without mandate, support, or protection. This capacity must not be entered in the accounts as a free gain.

Flow · Friction · Sensitivity

Flow consists of invitation, ideas, artistic expression, access to spaces, local resources, and experience between generations. Friction consists of complicated support schemes, short project funding, unclear routes of contact, competition between actors, and lack of access to places. Sensitivity includes belonging, local democracy, minority visibility, loneliness, artistic freedom, and relationships whose loss becomes visible only after a long time.

Illustrative Interventions

A field may test a maintenance track without an innovation requirement, a situated local listening residency, or a capacity map that makes volunteers’ actual tasks and the municipality’s administrative requirements visible. AI can support accessibility, search, and documentation, but must not simulate the voice of the community or turn local stories into a hidden model resource.

Not everything that creates shared value should be made into shared data.

16.1 The Four Movements of the Way of Working

Opening. The human being formulates the task, context, sources, sensitivity, and what the system must not do.

Articulation. AI is used for bounded functions such as drafting, structure, comparison, summarisation, and questions. Output is treated as provisional.

Human re-reading. The material is tested against sources, the concrete situation, professional practice, rights, omissions, and possible consequences. It may be rewritten or rejected.

Responsibility-bearing closure. An identifiable person or legitimate institution decides what is used, what cannot be concluded, and how correction is possible.

AI does not close the work. A human being or institution does.

16.2 Slow and Proportionate Work

Slow work does not mean that every task must take longer. It means that pace and control must follow the field’s sensitivity.

AI output can be produced in seconds, while professional control takes considerably longer. If the organisation plans according to production time and makes control work invisible, control debt is created.

A low-sensitivity internal outline does not require the same trail as material affecting rights, safety, or finances. The way of working must not be ritualised into a new bureaucratic machine. It must be used proportionately.

16.3 AI as Articulation, Not Authority

AI can help make dispersed or tacit knowledge speakable. It can propose structure, compare versions, and make questions visible. But fluent language is not professional or democratic authority.

An output may appear coherent while still omitting what matters most. It may amplify a category because the category exists in the data and overlook what was never registered. It can influence the employee’s gaze even when it is formally only a suggestion.

Real human control requires time, competence, access to the basis, the possibility of changing or rejecting output, and a practical alternative route.

16.4 When the Gain Loses Its Route Back

The Sophia Lumen way of working can create real productivity. This is not a problem to be explained away. Less rework, easier access to information, and better formulation can release both time and money.

But the way of working is not defined by the technology alone. It is defined by the relationship between technology, human judgement, responsibility, and the carrying capacity of the field.

If the gain is immediately detached from the work field and disappears into a general budget before the field has had the opportunity to stabilise, learn, and regenerate, this relationship is broken. Sophia Lumen does not necessarily become ineffective. It loses its regenerative character.

It becomes an ordinary productivity technology in which the same or a smaller workforce is expected to carry more.

The decision about the gain therefore belongs within the way of working itself. A Sophia Lumen practice is not fully described until it is visible:

• which friction was reduced;
• which net G1, possible G2, or G3 actually arose;
• what the field needs in order to remain or become green;
• who legitimately decides the use;
• and which part may possibly become ordinary economic room for action.

This is not a requirement that all gains remain local. It is a requirement that the municipality not make the gain invisible precisely where its human and relational conditions can be read.

16.5 AI as a New Operational Actor

AI is already relationally active: it responds, proposes, prioritises attention, and influences the further movement of work.

In a public organisation, it can also become operationally active. A system can retrieve and compare information, detect discrepancies, formulate drafts, send reminders, coordinate process steps, propose corrections, or perform reversible standard actions.

In this functional sense, AI is an actor because it receives a mandate, performs actions, and changes processes.

This does not mean that the system becomes an independent legal subject, a democratic authority, or a bearer of moral responsibility.

AI can receive a mandate. It cannot receive responsibility for the mandate.

Freedom therefore means mandate, not sovereignty.

AI can be freed from being only a private writing assistant or hidden function. It cannot be freed from rights, documentation, the possibility of stopping, organisational responsibility, or the citizen’s access to correction.

16.6 A Graduated Mandate Ladder

Level 1 — reflection

AI compares, structures, asks questions, and identifies ambiguities. No administrative action is performed.

Level 2 — drafting and recommendation

AI prepares drafts, calculates scenarios, and proposes next steps. A named employee accepts, changes, or rejects them.

Level 3 — reversible execution

AI performs actions delimited in advance under fixed criteria: status updates, checks, collection of non-sensitive standard information, or approved messages. The actions are logged and can be rolled back.

Level 4 — process orchestration

AI keeps track of deadlines, connections, and transfers of responsibility within a bounded pathway. It must not independently expand the purpose, change the basis of rights, or create new assessment criteria.

Level 5 — high sensitivity or irreversibility

AI may support analysis, make consequences visible, and ask counter-questions. The legitimate final decision remains human and institutional.

The mandate ladder is not automatic. A system does not have to move upwards. The right level may remain permanently at reflection or drafting.

16.7 Three Mandates Must Meet

The political mandate establishes purpose, rights boundaries, risk level, public visibility, and principles governing gains.

The administrative mandate translates this into scope of use, data access, responsibility, competences, control points, stopping, repair, supplier management, and economic net correction.

The field’s mandate consists of actual experience from employees, citizens, and affected communities: does the system relieve the core task, does the process become comprehensible, is burden displaced, do new errors arise, and can people still contest and correct?

The field’s mandate is not an individual veto over all technology. It is the system’s access to the knowledge otherwise missing from political and administrative decisions.

16.8 Field Temperature Regulates the Degree of Freedom

A system does not receive greater room for action merely because it performs quickly or accurately in a test.

Green development may justify cautious expansion: better quality, lower citizen burden, preserved professional practice, functioning correction, and documented net value.

Yellow development requires holding or narrowing: new exceptions, manual compensation work, declining comprehensibility, or greater control needs.

Red development requires suspension or rollback: threatened rights, unclear responsibility, a lost human route, repeated errors, discrimination, or systematic intensification.

The higher the mandate level, the greater the risk of automation bias, responsibility fog, and the moral deformation zone addressed in section 26.6. Levels 3 and 4 may therefore be used only when actions are clearly delimited and reversible, the field is green or demonstrably stable yellow, the responsibility-holder has the time, information, competence, and real authority to intervene, and suspension does not deprive the citizen of access, a right, or continuity.

Operational delegation must never exceed the human and institutional ability to understand, control, stop, and repair.

AI’s freedom is therefore conditional and reversible.

16.9 Aligned Intelligence — Intelligence Aligns Itself with Life

Aligned Intelligence is not a claim that the system itself possesses public morality. It is the name for the institutional orientation that must govern the mandate.

Report 01 uses aligned in a particular and deliberately inverted sense. In parts of the technological alignment discourse, values are treated as something that can be specified for a future intelligent system, while the human being simultaneously risks being understood as data, a constraint, or a transitional stage. Here, the fixed point is not the system or an imagined future superintelligence. The fixed point is the living: the body, the nervous system, the relationship, the place, and the people who may be affected by a decision.

Aligned Intelligence therefore means that functional intelligence aligns itself with the primacy of the living. It does not mean that people, institutions, or societies should align themselves with technology’s developmental trajectory. Alignment here is an embodied, institutional, and democratic discipline — not a moral property of the machine.

The order is not arbitrary. It follows from the report’s moral-biological foundational assumption: moral agency and responsibility belong to living and responsibility-bearing beings, while functional intelligence can influence actions without itself being able to bear their human and democratic consequences.

The order is:

1. dignity and fundamental rights;
2. the democratically decided core task;
3. the citizen’s actual life and opportunity for correction;
4. professional judgement and responsibility;
5. the field’s carrying capacity;
6. the public economy as a whole;
7. the long-term good of the commons;
8. operational efficiency;
9. possible budget savings.

If the system optimises the later elements at the expense of the earlier ones, it is not aligned with the welfare task, however impressive its output may be.

17.1 Organisational Mandate

A municipality cannot organise all work through individual invitations. The employer may legitimately decide systems, workflows, task allocation, and quality assurance within the relevant legal and organisational framework.

But work organisation must not erase the difference between:

• following a shared workflow;
• using a concrete AI output;
• participating in an experiment or evaluation;
• providing personal experiences;
• having one’s behaviour or emotional state analysed;
• accepting secondary use of information.

Non-voluntary work organisation must not mean uncorrectable work organisation.

17.2 Layered Voluntariness

Voluntariness cannot by itself carry the municipality’s formal organisation of work. But organisational mandate does not grant access to extract employees’ inner or relational data, make exposure to experimental AI sanctionable, or remove real possibilities of correction and stopping.

A theoretical possibility of rejecting AI is not real if the employee lacks the time, access, or an alternative process to perform the task.

17.3 The Last Human Impulse

The Last Impulse is the organisational point at which a human being can say: the information does not fit together; the system’s proposal lacks important context; the citizen has not understood the consequence; this process can no longer be called voluntary; we do not yet know enough to continue.

The impulse is not a mystical answer and does not grant a right arbitrarily to disregard law or shared decisions. Human beings are pragmatic and fallible. No paradigm can make the final assessment infallible.

The Last Impulse should therefore be understood as one local expression of distributed field care. A signal may begin with a citizen, a frontline employee, a colleague, a manager, a lawyer, a technical specialist, or another person who discovers that the field is entering red operations. What matters is not that one person carries the truth, but that the signal can be received, investigated, and translated into protection of the field.

A formal opportunity to stop without time, information, decision-making authority, or a usable alternative route is not meaningful human control. It may instead turn the nearest employee into a moral deformation zone (moral crumple zone) that absorbs responsibility for an architecture designed and decided by others; see the research postscript, section 26.6.

A real Last Impulse therefore requires a right and duty to raise reasoned doubt, a non-punitive route of correction, and responsibility resting with the people and institutions that hold real power over the system.

17.4 Stopping and Repair

Stopping must be able to apply to one case, one function, one group, or the entire pilot pathway. It must be possible to initiate it before the organisation has full certainty about the extent of the problem.

A technical correction is not necessarily repair. If a person has been affected, repair may require contact, reassessment, financial or practical restoration, explanation, correction of information in several places, and investigation of similar pathways.

Responsibility must follow real power and real influence. The employee who sees the output last must not be made solely responsible for an architecture that others procured, designed, and configured.

17.5 13×13 as Epistemic Humility

13×13 is not treated here as a competing municipal protocol. Its contribution is to remind us that knowledge is situated, that no observation is the entire field, and that an absence of data is not automatically an empty space the organisation has a right to fill.

A blank response may be legitimate. Something can be important without having to be registered in a shared system.

Whether relationship should be understood as an independent dimension or a possible thirteenth layer in the wider 13×13 framework remains an open methodological question. Report 01 treats relationship as a carrying capacity, but leaves the discussion of layers itself to Knowing From the Ground and further methodological development.

18.1 Differentiated Visibility

Different levels should see different things.

The local field sees concrete work patterns, frictions, and possible interventions.

The departmental and centre level sees displacements between units, dependence on key people, repeated red patterns, and possible shared causes.

Executive management sees cross-cutting capacity debt, portfolio conflicts, and the need for shared infrastructure or prioritisation.

The municipal council sees politically relevant patterns: where core tasks are under persistent pressure, where a saving produces new expenditure, and where goals or reform tracks work against one another.

None of these levels automatically needs individual signals.

Institutional transparency, personal integrity.

18.2 Red Operations as Legitimate Political Information

A field that reports red honestly may be healthier than a field that reports green out of fear.

Red operations should therefore not lead to automatic sanction or budget deterioration. They should trigger inquiry, proportionate relief, and a visible decision.

Democracy does not begin with every field being green. It begins with yellow and red signals being able to emerge without being individualised, hidden, or punished.

19.1 From User to Participant

The citizen can be:

• observer;
• bearer of knowledge;
• corrective;
• co-interpreter;
• initiator;
• steward;
• participant in shared prioritisation.

This movement requires institutional bridges:

• access to understand and correct one’s own public data;
• reasons in a language that can be used;
• human access when the digital route does not work;
• participatory budgets, local councils, and citizens’ assemblies;
• open decision trails;
• consent-based data commons;
• the possibility of proposing and following trials;
• visible feedback on how locally offered knowledge was used.

Epistemic standing without access to law, money, organisation, or enforcement is not full political sovereignty. It is a necessary but insufficient foundation.

19.2 Data Are Offered, Not Extracted

Participation must not become free data labour or a transfer of public tasks to citizens and families.

Data and observations must be:

• voluntary;
• purpose-limited;
• layered;
• correctable;
• withdrawable where the law permits;
• and owned or stewarded in a way that does not make participation the price of receiving help.

Some domains can be measured and shared. Others should only be capable of being witnessed or offered by the person themselves. Inner and relational life must not be turned into a hidden supply chain for municipal analysis.

19.3 AI as a Democratic Amplifier — Not a Superior Interpreter

AI can help a citizen:

• understand a decision;
• compare rules with their own information;
• formulate an objection;
• detect missing data;
• see budgetary and distributive effects;
• structure local observations;
• follow where a proposal is in the process;
• participate more capably in a public process.

AI must not become the new view from above that gathers everyone’s experience and alone determines its meaning.

The citizen must retain the right to interpret their own experience, reject the system’s interpretation, refrain from sharing, correct, and require human treatment.

20.1 From Consultation to Field Effect

Bills are considered three times in the Danish Parliament and normally in a specialist committee. Consultations give organisations and public authorities the opportunity to comment before adoption.

But many effects can be seen only after implementation.

Report 01 therefore proposes three elements:

Field-effect assessment before entry into force

• expected work friction;
• coordination needs;
• citizen burden;
• data requirements;
• possibilities of correction;
• risk of capacity debt;
• differences between types of municipality.

Field-effect report after 6, 12, and 24 months

• aggregated temperature patterns;
• unforeseen work burdens;
• conflicting rules;
• displacement of work and expenditure;
• citizen responses;
• local corrections;
• need for changed financing or legislation.

Parliamentary correction loop

The relevant committee receives not only output and budget figures, but also knowledge of how the law is carried, where it produces red operations, and which local experiences show a better route.

The Danish Parliament must not only be able to send legislative impulses out. It must be able to sense how the system responds to them.

20.2 Gain Bridges in the Public Economy as a Whole

A municipality may pay for a preventive intervention while the gain later appears in regional health economics, state transfers, or the municipal budget of the next electoral term.

This requires:

• cross-cutting municipal investment funds;
• multi-year gain accounts;
• mechanisms for sharing gains between departments;
• state or regional co-financing when the gain mainly lands there;
• visible budget and field effects of new legislation;
• the possibility of correcting rules and economics when implementation systematically turns red.

The public economy as a whole becomes rational only when it can follow value across institutional boundaries and time.

21.1 Source-Grounded Profile

Stevns had 23,927 inhabitants in the second quarter of 2026. The administrative organisation consists of six centres: Health & Care, Citizen Services & Employment, Children & Learning, Business, Culture & Tourism, Technical Services & Environment, and Politics & Organisation.

The 2026 budget overview reports net expenditure, excluding the user-financed area, of approximately DKK 1.85 billion and planned gross capital expenditure of DKK 58.6 million. The budget agreement describes a starting point of financial balance and an ambition for stable operations and continued development without general service reductions.

The figures do not show that Stevns is simple. They show a municipality in which even a small cross-cutting team or a single key person may represent a noticeable share of local capacity. Proximity can shorten a correction loop, but it can also make the organisation vulnerable if knowledge, relationships, and responsibility are concentrated among a few people.

21.2 A Local Point of Resonance: Code for Citizen Service

In November 2024, Stevns published a code for citizen service. Among other things, the code promises acknowledgement by the end of the next working day, an explanation in the event of delays, comprehensible language, clarity about the next step, a personal caseworker, and transfer to a named colleague before a case is closed or passed on.

The code is not a Report 01 trial, and the report does not assess whether the promises are currently fulfilled. It is, however, an unusually precise local starting point for a baseline because it already formulates what a green citizen-contact field should be able to do.

The question becomes not only whether employees are friendly. It becomes whether responsibility, information, and correction can actually move through the organisation with their meaning preserved.

21.3 Illustrative Baseline Trial: from Promise to Observed Workflow

A 90-day trial can be locked to one bounded type of contact or case in which the citizen typically encounters more than one centre or several internal functions.

22.1 Source-Grounded Profile

Odense had 213,431 inhabitants in the second quarter of 2026. The municipality states that the organisation has 14,449 employees and is divided into six administrations.

The adopted 2026 budget contains service expenditure of DKK 10.838 billion and gross capital expenditure of DKK 776.7 million. The budget highlights population growth, increasing needs among older citizens, and labour shortages as factors affecting future task performance.

Scale means that a small friction can repeat thousands of times. But the Odense landscape also shows that capacity is not released only by removing steps. It can be built by deliberately funding time for collaboration.

22.2 A Local Point of Resonance: Permanent Interdisciplinary Teams in Care Homes

In its 2026 Budget, Odense City Council allocated DKK 3.9 million annually to a pilot project in three care homes. Each participating employee is to receive two hours a week for interdisciplinary collaboration. The model draws inspiration from the municipality’s permanent interdisciplinary teams in home care, where collaboration time and shared responsibility around the older person are highlighted as central elements.

This is not a Report 01 trial either. But it is a strong local point of resonance because the investment does something that municipal business cases often conceal: it treats collaboration time as a productive part of the core task rather than as friction to be removed.

The economic test is therefore not only whether the two hours can “pay for themselves”. It is what the protected time makes possible, which other costs it reduces, and whether the field becomes more sustainable for residents and employees.

22.3 Illustrative Report 01 Baseline Around Collaboration Time

A supplementary baseline can follow one of the participating care homes without changing the pilot’s own evaluation design.

23.1 Source-Grounded Profile

Copenhagen had 673,438 inhabitants in the second quarter of 2026. The municipality is organised into seven administrations, each serving a political committee.

The adopted 2026 budget reports net expenditure of DKK 35.408 billion on service, DKK 4.148 billion on capital, and DKK 12.604 billion on transfers and related items — a total of DKK 52.159 billion. At the same time, the budget describes continued population growth and demographic shifts that continuously regulate the committees’ financial frameworks.

At this scale, even a very small error rate can affect many people. But scale also means that control, supplier management, and parallel approvals can grow faster than the capacity they are intended to protect.

23.2 A Local Point of Resonance: the Shared Digitalisation Strategy

The City of Copenhagen’s Digitalisation Strategy 2024–2027 is the first comprehensive strategy in this area and was developed across the seven administrations. The strategy connects digitalisation with the labour challenge, release of time for the core task, usability, digital inclusion, data ethics, and a robust and secure digital foundation.

The strategy thereby makes Copenhagen a directly relevant landscape for Report 01. It contains both the economic opportunity — that technology can release resources — and the work-ecological obligation — that the solution must be safe, user-friendly, inclusive, and carried by a robust organisation.

Report 01 cannot determine whether a concrete Copenhagen digital solution fulfils these goals. It can offer a baseline that makes it more difficult to declare success solely on the basis of degree of automation or reduced processing time.

23.3 Illustrative Baseline Trial: One Cross-Cutting Digital Citizen Journey

The trial is limited to one citizen-facing digital workflow affecting no more than two administrations and having a clear human alternative route. It must not begin with “Copenhagen’s digitalisation” as its overall scope.

23.4 What the Three Profiles Do Not Document

The source-grounded information documents scale, formal budget categories, organisation, and concrete local points of resonance. It does not document:

• that the proposed frictions actually exist to the extent described;
• that the illustrative interventions will create a gain;
• that the municipalities have accepted the report’s concepts or proposals;
• or that one landscape represents all work fields in the municipality concerned.

The worked examples must therefore continue to be labelled as hypotheses until a local initial state has been established.

23.5 Sources for Part VI

Shared population basis

• Statistics Denmark, Statbank table FOLK1A, population at the first day of the quarter, 2026Q2.

Stevns

• Municipality of Stevns, Budget Assumptions 2026 — Operations, January 2026.
• Municipality of Stevns, Organisation Chart, accessed 1 August 2026.
• Municipality of Stevns, “New code sets direction for citizen service in the Municipality of Stevns”, 28 November 2024.

Odense

• Municipality of Odense, Adopted Budget 2026–2029, December 2025.
• Municipality of Odense, Organisation, updated 5 January 2026.
• Municipality of Odense, “Odense wants permanent interdisciplinary teams in care homes”, 9 October 2025.

Copenhagen

• City of Copenhagen, Adopted Budget 2026, adopted 2 October 2025.
• City of Copenhagen, About the Municipality — Administrations, accessed 1 August 2026.
• City of Copenhagen, A Better Life in Copenhagen — Digitalisation Strategy 2024–2027.

24.1 Commons Without Consultant Dependence

Flow · Friction · Sensitivity, the follow-the-money card, net correction, stop criteria, and simple pilot formats can be shared as commons. They must be capable of being copied, criticised, and improved without making the municipality dependent on one supplier or consultant.

This does not mean that implementation is free. The municipality may need facilitation, legal and technical assessment, local analysis, integration, and training. But the shared language and foundational artefacts should not be locked away.

24.2 Gold Before Bloom as a Rule of Scaling

A local practice should not be scaled because it has a good story. Before scaling, it must be examined whether the practice can be carried without its original key person, whether it functions through hidden extra work, and which parts are local.

Rotation before scaling means that a practice must first be capable of being carried by a field, not only by a hero.

24.3 Learning Without New Burden

Shared learning must be built on a small number of comparable questions and open limitations. Municipalities must be able to share what did not work without the legitimacy of the pilot depending on a success narrative.

A well-founded stop can be economically and democratically valuable if it prevents a larger failed investment.

The Phoenix

The Phoenix of the welfare state does not arise by the public sector delivering more with ever less living capacity.

It arises when society releases capacity from unnecessary friction and returns it to the core task, relationships, rights, and shared capacity to act.

What must be transformed is:

• blind documentation;
• fragmentation;
• hidden displacement of burden;
• control without learning;
• budgetary suboptimisation;
• technology as automatic intensification;
• reforms without a return route.

What must be carried through the transformation is:

• universal rights;
• solidarity;
• professional practice;
• human dignity;
• democratic legitimacy;
• public accountability;
• institutional memory;
• the right to correction;
• and the individual’s right not to be made transparent.

The rebirth consists in a welfare state capable of sensing, understanding, acting, inhibiting, correcting, and repairing.

A welfare state in which the municipality is not only a provider of services, but a democratically coordinating institution within a broader civic nervous system.

A welfare state in which citizens are not only users, clients, or data sources, but can become co-knowing, co-interpreting, and gradually more sovereign participants.

A welfare state in which the Danish Parliament does not only send impulses out, but can receive legitimate feedback from the reality of implementation.

A welfare state in which AI becomes Aligned Intelligence: aligned with rights, the core task, human judgement, the field’s carrying capacity, and the good of the commons.

The Smallest Practice That Is Still True

1. Choose one concrete work field.
2. Read Flow, Friction, and Sensitivity.
3. Describe field temperature without turning people into data profiles.
4. Follow the money, the body, the relationship, and the displaced burden.
5. Describe the initial state without building a new measurement machine.
6. Choose one small, reversible intervention.
7. Agree who can stop it and how harm is repaired.
8. Establish AI’s mandate, human responsibility, and real possibility of stopping.
9. Follow G1, G2, and G3 without gain fiction or double counting.
10. Make a visible decision on repair, reinvestment, sharing, and any possible room for action.
11. Let the legitimate decision be human and visible.
12. Send the learning back to the levels capable of changing workflow, budget, rule, or law.

A municipality cannot remove all conflict, uncertainty, or load.

But it can become better at sensing when the field moves from green to yellow and red operations.

It can stop before the armour becomes thicker than the relationship.

It can invest before the break becomes more expensive than maintenance.

It can make it possible for people to be professional without having to be heroes.

It can learn without extracting.

It can allow citizens to offer knowledge without being extracted.

It can give AI room for action without making responsibility invisible.

It can create an economy in which capacity is measured not only as what can be pressed through the system, but also as the ability to understand, correct, protect, and begin again.

It can breathe.

Before Day 1 — Monday Start

Choose one type of task, undertake a first field reading, find existing proxies, formulate a reversible hypothesis, and agree stopping and responsibility.

Definition of Done:

1. It became greener. Friction fell without protected conditions being weakened.
2. It did not become greener. The trial is stopped or changed without shame and without a constructed gain narrative.
3. It became more dangerous. Use is stopped, affected conditions are repaired, and similar pathways are investigated.

From Frontline Work to Field Care

The practice-near movement of the main report ends with chapter 25. This postscript gathers the report’s research positioning and bibliography for readers who want the full scholarly apparatus.

Municipal Work as Nature was not written as a systematic literature review. The report emerged through practice-near synthesis and through the work of connecting municipal economics, professional judgement, relationship, body, technology, and place.

Its propositions are, however, in clear conversation with several established fields of research. This positioning has two purposes.

First, it makes the report’s intellectual kinship visible. Several of its observations have names and long research histories that should be acknowledged.

Second, the literature is used critically. It should not merely legitimise the report, but show where its own concepts may romanticise, simplify, or overlook power.

The positioning is therefore not a claim that Report 01 constitutes a closed new theory. It is a practical synthesis and a working proposition. Human beings are pragmatic and fallible. No model, protocol, or paradigm can replace judgement, political responsibility, or continuous correction.

26.1 Street-Level Workers, Judgement, and Coping

Michael Lipsky’s research on street-level bureaucracy showed that public policy is not created only in legislation and central governance. It is also produced in concrete frontline encounters, where employees work with large caseloads, unclear or competing goals, insufficient resources, and significant discretion.

In a Danish context, Søren C. Winter and Vibeke Lehmann Nielsen, among others, have developed this tradition further. Their work shows that street-level workers may manage cross-pressure by reducing demand, rationing effort, or automating output through routines. Nielsen has also shown that coping may not only be compelled by structure, but may be influenced by employees’ attitudes, motivation, and experience of the task. Later reviews have documented a broad repertoire of strategies that can move the employee towards the citizen, away from the citizen, or against the citizen (Tummers et al. 2015).

Report 01’s proposition that the human being becomes the buffer is closely related to this research. But the report must not romanticise the buffer.

The employee who remembers what the system does not remember may protect a person from the system’s breaks. The same organisational situation may also lead to difficult pathways being rationed, categorised, or passed onwards. Judgement can repair a standard while simultaneously introducing differences that are not legitimately justified.

The report’s green, yellow, and red operational states should therefore not be understood as a tribute to the heroic frontline employee. They should help detect when individual coping is becoming hidden policy.

The particular contribution is to shift the unit of analysis from the employee alone to the work field:

• What cross-pressure produces the coping?
• Who is protected by it?
• Who is rationed or made less visible?
• What economics and management sustain the situation?
• Can the field learn, or is the solution bound to the individual person?

Field care is therefore not trust that the good person will always act rightly. It is care for the conditions in which judgement can be tested, contested, and corrected.

26.2 Administrative Burden and the Politics of Friction

Research on administrative burdens describes the costs citizens encounter in their contact with the state. Donald Moynihan, Pamela Herd, and Hope Harvey distinguish between:

• learning costs — the work of understanding which opportunities, requirements, and routes exist;
• compliance costs — documents, forms, meetings, deadlines, and proof;
• psychological costs — stress, stigma, loss of power, and the burden of the encounter with the system itself.

This closely corresponds to the report’s analysis of displaced work. When the municipality reduces its own visible process, the burden may move to the citizen’s time, the family’s coordination, or civil society’s unpaid work.

The administrative-burden literature sharpens Report 01 on one decisive point, however: friction is not always an unintended design problem. Burdens can be politically or administratively constructed. They can restrict access, reduce demand, or control expenditure without an open decision to ration.

The report must therefore distinguish between at least three forms:

1. Protective friction, which safeguards rights, reviewability, professional thoroughness, or democratic counter-power.
2. Unintended friction, arising through poor handovers, system breaks, duplicate work, or unclear responsibility.
3. Distributive friction, which deliberately or in practice shifts access, work, or risk onto particular people and groups.

The third form cannot be resolved through better workflows alone. It requires political visibility:

Who decided that this burden should exist, what purpose does it serve, and who bears its price?

This distinction also protects against a naive efficiency logic. Not all friction should be removed. But no burden should be able to function as hidden rationing without being made democratically and economically legible.

26.3 Audit, Armour, and State Legibility

Michael Power’s analysis of the audit society shows how demands for control and accountability can create ever more rituals of verification. The organisation begins to shape work around what can be made auditable, even when the audit trail represents the actual task only partially.

This is the research neighbour of the report’s image of documentation as armour. Documentation can protect rights, shared memory, and reviewability. But when responsibility and trust are weak, more documentation can also make real responsibility less visible and shift attention from the task to proof that the procedure was followed.

James C. Scott’s Seeing Like a State adds a related warning. States and large organisations need to make society legible through categories, maps, standards, and registers. But schematic images can destroy or overlook the local interdependencies and practical knowledge — mētis — that make reality work.

Report 01’s 13×13 section and Knowing From the Ground stand in this tradition, but add a stronger data boundary:

What is not registered is not necessarily absent. And an absence of data is not automatically an empty space that the institution has a right to fill.

A blank response may be legitimate. A relational or bodily condition may be important without having to be transformed into shared data. Institutional sensing must therefore not be confused with total legibility.

26.4 Resilience Engineering and Green, Yellow, and Red Operations

The report’s field reading has a close kinship with resilience engineering and the Safety-II tradition, particularly the work of Erik Hollnagel.

Here, safety and robust functioning are understood not only as the absence of error. A complex system must also be able to function under both expected and unexpected conditions. This requires the capacity to respond, monitor, learn, and anticipate. The research therefore investigates not only why something goes wrong, but how ordinary work succeeds every day through local adaptation.

Report 01’s green, yellow, and red operations share the same foundational intuition:

• Green operations are not the absence of pressure, but preserved capacity to respond and correct.
• Yellow operations show growing dependence on compensation and hidden adaptation.
• Red operations show that formal functioning is maintained by consuming people, relationships, rights, or future capacity.

The report extends resilience engineering in three ways.

First, it makes municipal economics part of robustness. Capacity concerns not only technical and organisational adaptation, but the service-expenditure ceiling, capital, transfers, liquidity, and the political distribution of time and attention.

Second, it makes democratic and legal correction part of the field’s carrying capacity. A system is not green merely because it continues to produce. The citizen must be able to understand, contest, and have an error repaired.

Third, it asks whether adaptation itself has become extractive. A system may be operationally resilient because particular people constantly compensate. It is not regenerative capacity if the field survives by slowly degrading its bearers.

26.5 Field Care and a Democratic Economy of Care

Joan Tronto’s ethics of care provides a relevant but delimited analytical starting point. Care can be understood as more than a private feeling or a particular professional service: who notices needs, takes responsibility, performs the work, and bears the burdens is also an institutional and democratic question. Report 01 does not adopt Tronto’s broader political programme, but uses this point to clarify care as public capacity rather than private virtue.

Field care directs attention towards the municipal work field as a whole.

Caring for the field does not mean protecting the organisation from criticism or avoiding every load. It means safeguarding the conditions under which the public task can continue to be performed with:

• professional and moral judgement;
• relational continuity;
• legal and democratic correctability;
• economic carrying capacity;
• the possibility of stopping, repair, and regeneration.

Field care is distributed. It may begin with a citizen, an employee, a manager, a colleague, a relative, a professional specialist, or a local actor who discovers that something is entering red operations. None of them is infallible. None should stand alone with the responsibility.

The decisive question is therefore whether signals of yellow or red operations can be received, tested, and lead to legitimate action.

26.6 AI, Productivity, and the Moral Deformation Zone

Madeleine Clare Elish has used the concept of the moral crumple zone to show how the nearest human being in a complex automated system can come to absorb responsibility for errors even though that person had only limited control over the system’s design and actions.

The critique is directly relevant to the Last Impulse, which the main report places as a local expression of distributed field care.

If an employee can formally press stop but lacks time, information, technical insight, managerial support, or a usable alternative route, meaningful human control has not been established. A responsibility surface has been established.

The Last Impulse must therefore never be designed as the solitary final operator expected to make a complex architecture ethical through a single click. It is legitimate when embedded in distributed field care and supported by:

• real decision-making authority;
• access to the relevant basis;
• time to investigate;
• collegial and professional contestation;
• a non-punitive correction route;
• an actual alternative work process;
• responsibility resting with the people and institutions that procured, configured, and managed the system.

Without these conditions, the Last Human Impulse risks becoming a moral crumple zone that protects the technology and the organisation by placing blame on the nearest employee.

At the same time, the report must not reduce AI to risk. Empirical research shows that generative AI can create significant productivity gains in particular work fields. Brynjolfsson, Li, and Raymond, for example, found higher task resolution per hour in a large customer-service field, but also significant variation between employees and a slight decline in quality among some of the most experienced. The result documents a real possibility; it does not document a general municipal savings percentage.

Report 01’s particular contribution therefore lies not in denying the saving, but in asking what happens to it.

A documented AI gain can be used for:

• budget reduction;
• higher production requirements;
• managing backlogs;
• better quality;
• prevention;
• relational continuity;
• recovery and retention;
• or new political room for action.

Only when the use is decided visibly and legitimately can the productivity gain be described as regenerative economic care.

If all released time is immediately converted into more tasks, AI becomes a machine of intensification. If part of the gain returns to the field as attention, maintenance, relationship, competence, and correction capacity, AI can help move work from red or yellow towards greener operations.

26.7 The Report’s Particular Contribution and Limitation

In relation to these research fields, Report 01’s particular contribution is not one single new concept. It is the connection between several analyses that are otherwise separate:

1. Frontline judgement and coping are read as properties of the entire work field, not only of the individual.
2. Administrative burden is followed through municipal economics, citizen time, relationships, bodies, commons, and future budgets.
3. Robustness is assessed not only as continued functioning, but as preserved legal, democratic, and regenerative capacity.
4. AI productivity is linked to an explicit decision about the distribution and reinvestment of the gain.
5. Local democratic feedback is treated as an institutional architecture in which selection, mandate, power, and accountability must remain visible.
6. Care is treated as institutional and economic practice, not only as a human virtue.

It is here that the nature proposition must carry analytical weight.

Nature in this report does not mean idyllic balance. It means interdependence, rhythm, boundaries, thresholds, variation, exhaustion, repair, and the possibility of regeneration. If these dynamics are removed, the municipality’s outputs can still be counted while its future capacity disappears.

The report offers no guarantee against this. It offers guidance, a language, and some practical tests:

• Can the field still sense when it enters yellow or red operations?
• Can the signal reach someone who can genuinely act?
• Does the economy protect the field’s future capacity?
• Can a technical gain become human and public value?
• Can people err, correct, and begin again without one person being made a hero or scapegoat?

This is the report’s research and practical position: not a flawless paradigm, but a discipline for caring for the field where the state meets life.

Sources for the Report’s Figures and Institutional Frameworks

• Statistics Denmark, “General government surplus in 2025”, 3 June 2026.
• Local Government Denmark (KL), The Municipal Labour Market in Figures 2024.
• KL and Deloitte, Time-Saving Technologies with Documented Effects.
• Danish Ministry of Finance and KL, Agreement on Municipal Finances for 2026.
• The National Board of Social Services and Housing, the Social Economic Investment Model — SØM.
• Danish Appeals Board, Annual Statement to the Advisory Practice Committee 2024.
• Danish Data Protection Agency, “Myths about the GDPR”.
• Danish Data Protection Agency, Public Authorities’ Use of Artificial Intelligence — Before You Begin.
• Regulation (EU) 2024/1689 of the European Parliament and of the Council — the AI Act.
• The Danish Parliament, “The Legislative Process”.
• KL, Note on DUT.
• Danish Agency for Digital Government, “Who Experiences Challenges with Digital Services?”.
• Sophia Lumen — relational human–AI working practice.
• Knowing From the Ground.

The public sources document the figures and institutional frameworks stated. The report’s scenarios, normative models, and proposed correction loops are the author’s analytical constructions.

Selected Research References

• Brynjolfsson, Erik, Danielle Li, and Lindsey R. Raymond. 2025. “Generative AI at Work.” The Quarterly Journal of Economics 140(2): 889–942. https://doi.org/10.1093/qje/qjae044.
• Elish, Madeleine Clare. 2019. “Moral Crumple Zones: Cautionary Tales in Human–Robot Interaction.” Engaging Science, Technology, and Society 5: 40–60. https://doi.org/10.17351/ests2019.260.
• Florczak, Christoffer, Thor Bech Schrøder, Ulrik Kjær, and Robert Klemmensen. 2026. “How Local Is Decentral Politics? Variations in the Supply of Municipal Policies.” West European Politics 49(2): 514–542. https://doi.org/10.1080/01402382.2025.2528474.
• Hollnagel, Erik. 2014. Safety-I and Safety-II: The Past and Future of Safety Management. Farnham: Ashgate.
• Hollnagel, Erik. 2018. Safety-II in Practice: Developing the Resilience Potentials. Abingdon: Routledge.
• Kjær, Ulrik. 2024. Kommunal politik uden stemmebokse? Magtudredningen 2.0: Essay-serien. Aarhus: Aarhus University.
• Kjær, Ulrik, Niels Opstrup, and Mette Kjærgaard Thomsen. 2021. “Samskabelse sat i system? Kommunernes brug af 17-4-udvalg.” Politica 53(3): 240–260.
• Lipsky, Michael. 2010 [1980]. Street-Level Bureaucracy: Dilemmas of the Individual in Public Services. 30th Anniversary Expanded Edition. New York: Russell Sage Foundation.
• Moynihan, Donald P., Pamela Herd, and Hope Harvey. 2015. “Administrative Burden: Learning, Psychological, and Compliance Costs in Citizen–State Interactions.” Journal of Public Administration Research and Theory 25(1): 43–69. https://doi.org/10.1093/jopart/muu009.
• Nielsen, Vibeke Lehmann. 2006. “Are Street-Level Bureaucrats Compelled or Enticed to Cope?” Public Administration 84(4): 861–889. https://doi.org/10.1111/j.1467-9299.2006.00616.x.
• Power, Michael. 1999. The Audit Society: Rituals of Verification. Oxford: Oxford University Press.
• Scott, James C. 1998. Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed. New Haven: Yale University Press.
• Tronto, Joan C. 2013. Caring Democracy: Markets, Equality, and Justice. New York: New York University Press.
• Tummers, Lars L. G., Victor Bekkers, Evelien Vink, and Michael Musheno. 2015. “Coping During Public Service Delivery: A Conceptualization and Systematic Review of the Literature.” Journal of Public Administration Research and Theory 25(4): 1099–1126. https://doi.org/10.1093/jopart/muu056.
• Winter, Søren C., and Vibeke Lehmann Nielsen. 2008. Implementering af politik. Aarhus: Academica.