AIFC-002: Community Model

Status: Draft 0.1 Standard: AI-First Community Standard Short name: AIFC Related to:

• AIFC-000: Manifesto for AI-First Communities
• AIFC-001: Core Concepts

Purpose of this document: Describe the basic model of an AIFC community: how purpose, values, knowledge, decisions, work, learning, feedback, AI involvement, and interfaces with other communities fit together.

1. Purpose of this document

This document defines the basic architecture of an AIFC community.

AIFC-000 explains why the standard exists. AIFC-001 defines the core terms. AIFC-002 describes how those terms form a functioning system.

The goal is to show that an AIFC community is not only:

• a team,
• an organizational unit,
• a knowledge base,
• a workflow,
• an AI agent configuration,
• or software.

An AIFC community is a living system with purpose, values, knowledge, work, decisions, feedback, and responsibility.

2. Community as the primary unit

The primary unit of AIFC is not AI.

The primary unit is a community with purpose.

A community may be small or large:

• work team,
• project team,
• department,
• company,
• family,
• school,
• municipality,
• state,
• professional group,
• nonprofit organization,
• network of communities,
• or represented ecosystem.

AIFC assumes that each community has, or needs to have:

• purpose,
• values,
• boundaries,
• members or represented actors,
• knowledge,
• decisions,
• work,
• learning,
• a feedback mechanism,
• and an interface with other communities.

AI may help the community. AI may accelerate its work. AI may detect signals. AI may propose changes. But AI does not own the community.

The community owns its purpose.

3. The basic AIFC community pattern

An AIFC community consists of several basic layers:

Purpose
down
Values
down
Strategy / Direction
down
Knowledge Base
down
Decisions
down
Work / Execution
down
Learning / Retrospective
down
Feedback / Change Proposals
down
Updated Knowledge / Strategy / Values Interpretation

This model is not a one-way linear pyramid.

It is a cycle.

Purpose and values give work its direction. Work creates experience. Experience generates signals. Signals may trigger change proposals. Approved changes update the knowledge base, workflows, skills, strategy, or interpretation of values.

An AIFC community is therefore defined by two movements:

Top-down flow:
values -> purpose -> strategy -> work

Bottom-up flow:
experience -> signals -> change proposals -> decisions -> system updates

4. Top-down flow

The top-down flow is how the community translates its purpose into work.

4.1 Values

Values define what the community does not want to sacrifice.

They are not decoration. They are the highest governance layer.

Values answer questions such as:

• What matters to us?
• Which boundaries do we not want to cross?
• What must we not sacrifice under pressure for speed, performance, or profit?
• Which behavior is incompatible with our community?

4.2 Purpose

Purpose defines why the community exists and where it wants to go.

Purpose answers questions such as:

• Why do we exist?
• What value do we want to create?
• Who or what do we serve?
• Which desired state do we want to contribute to?

4.3 Strategy

Strategy translates purpose into a path.

It answers questions such as:

• How do we move from the current state to the desired state?
• Which priorities make sense now?
• Which trade-offs do we accept?
• What are we not doing?

4.4 Work

Work is the concrete execution of purpose.

It may take the form of:

• projects,
• epics,
• tasks,
• support activities,
• maintenance activities,
• experiments,
• decisions,
• communication,
• delivery,
• operations,
• care.

AIFC requires work to be traceably connected to purpose, values, and decisions.

5. Bottom-up flow

The bottom-up flow is how experience from reality returns to the system.

Without this movement, the community would execute a plan but would not learn.

The bottom-up flow contains:

Work / Execution
down
Experience
down
Observed Signals
down
Change Proposals
down
Decision
down
Update of Knowledge / Workflow / Strategy / Values Interpretation

5.1 Experience

Experience emerges from work.

It may show:

• what works,
• what does not work,
• where noise appears,
• where value is lost,
• where AI helps,
• where AI creates dependency,
• where a process harms values,
• where a new opportunity appears.

5.2 Observed Signal

An observed signal is a meaningful signal from reality.

It may come from:

• team work,
• customer feedback,
• support,
• incident,
• market,
• data,
• retrospective,
• another community,
• AI analysis,
• or represented ecosystem.

5.3 Change Proposal

A change proposal is a structured proposal for change.

It may be proposed by:

• a person,
• a team,
• a customer,
• another community,
• an AI agent,
• or an environmental representative based on a data signal.

A change proposal may target:

• workflow,
• skill,
• rule,
• priority,
• strategy,
• value interpretation,
• Community Interface,
• governance,
• or a higher level of the community.

5.4 Decision

A change proposal is not a decision.

Every significant proposal must be evaluated and decided by responsible human or community governance.

AI may formulate a proposal. AI may analyze impact. AI may recommend a next step. The community decides.

6. Knowledge layer

The knowledge layer is the community’s memory.

It contains:

• purpose,
• values,
• strategy,
• current state,
• desired state,
• path,
• decisions,
• principles,
• workflows,
• skills,
• retrospectives,
• change proposals,
• interfaces,
• AI governance,
• security rules.

The knowledge layer must be:

• human-readable,
• agent-actionable,
• software-verifiable,
• versioned,
• auditable,
• protected,
• usable without AI.

AIFC prefers textual, open, and versionable formats, such as Markdown with metadata in Git.

Not because Markdown or Git are the only possible technologies, but because they support three basic AIFC properties:

human-readable
agent-actionable
software-verifiable

The knowledge layer is the Source of Truth, but by itself it may not be accessible enough for every community member.

Markdown, metadata, Git history, and validation rules are suitable for AI agents, software, audit, and long-term know-how management. A person often needs an assisted layer that helps them quickly understand the state of the system, decide, and act.

AIFC therefore distinguishes:

• Source of Truth as the authoritative memory of the community,
• Human Cockpit Layer as the human-operable access layer to that memory.

Without a human access layer, even a high-quality knowledge base can become a system that agents and validators understand, but that the community finds too hard to operate.

7. Human Cockpit Layer

The Human Cockpit Layer is the human-operable layer over the community knowledge base.

Its purpose is to protect human attention and allow community members to manage the system without having to work directly with the full technical structure of the Source of Truth.

The Human Cockpit Layer may show:

• community purpose,
• values,
• strategy,
• current state,
• priorities,
• backlog,
• decisions,
• risks,
• skills,
• AI capacity,
• AI usage,
• AI waste,
• change proposals,
• pending approvals,
• the relationship between work, values, and purpose.

The Human Cockpit Layer is not necessarily one specific product. It may be an application, dashboard, documentation interface, assisted editor, workflow UI, or another way of making the Source of Truth accessible to people.

AIFC assumes that AI agents and software may work directly with the structured knowledge base. A person needs a layer that helps them quickly understand, decide, and act.

The Human Cockpit Layer is therefore not a replacement for the knowledge base.

It is human access to it.

An AIFC community should ensure that its know-how is not only machine-processable, but also human-operable.

8. Decision layer

The decision layer describes how the community decides.

Every significant decision should have:

• owner,
• context,
• reason,
• affected values,
• considered alternatives,
• expected impacts,
• decision level,
• Decision Record,
• review rules.

An AIFC community must distinguish:

• proposal,
• recommendation,
• decision,
• approved change,
• implemented change.

An AI-generated proposal must not be treated automatically as a decision.

8.1 Decision levels

A decision may have different levels:

local decision
team decision
department decision
company decision
owner / board decision
cross-community decision
higher-level governance decision

A change proposal must be routed to the correct decision level.

For example:

• a change to a local checklist may be a team decision,
• a change to a security rule may be a security governance decision,
• a change to a value interpretation may require leadership or community owner decision,
• a proposal affecting other communities must pass through the Community Interface.

9. Work layer

The work layer is where purpose becomes action.

AIFC distinguishes at least three types of work.

9.1 Development / change work

Work that moves the community into a new state.

Examples:

• new product,
• new process,
• new page,
• new service,
• architecture change,
• migration,
• transformation.

9.2 Maintenance work

Maintenance work keeps the system healthy.

Examples:

• documentation maintenance,
• knowledge base cleanup,
• refactoring,
• rule updates,
• metadata repair,
• skill maintenance,
• security maintenance,
• workflow maintenance,
• decision updates,
• outdated information review.

Maintenance work is not second-class work.

Everything the community does not care for tends to degrade or create debt.

This may create:

• knowledge debt,
• process debt,
• security debt,
• decision debt,
• skill debt,
• AI dependency debt,
• technical debt,
• or trust debt between communities.

Maintenance protects the community’s ability to continue moving toward its purpose.

AI may help significantly with maintenance work: finding outdated information, proposing cleanup, identifying duplication, flagging missing metadata, or detecting recurring problems.

AIFC requires repeated maintenance routines to be gradually converted into stable non-AI workflows where that makes sense.

9.3 Support work

Work that responds to needs, problems, or incidents.

Examples:

• support ticket,
• incident,
• customer question,
• internal request,
• bug,
• urgent operational problem.

AI may help in all types of work. AIFC requires clarity about:

• where AI helps,
• what its scope is,
• which data it uses,
• who owns the result,
• whether fallback exists,
• and whether know-how from the work returns to the Source of Truth.

10. Learning layer

The learning layer ensures that the community does not stay the same, but learns.

It contains:

• retrospectives,
• lessons learned,
• AI retrospective,
• Skill Evolution,
• workflow conversion,
• AI waste backlog,
• updates to the knowledge base,
• updates to human skills,
• updates to AI skills.

An AIFC community should ask:

• What did we learn?
• What should change?
• What should be written down?
• What should be converted into a workflow?
• What should be added to a skill?
• What should be escalated as a change proposal?
• What should AI no longer do repeatedly?
• Where have we lost human capability?
• Where is lock-in emerging?

The learning layer connects work with the future quality of the community.

11. Feedback layer

The feedback layer is broader than a retrospective.

A retrospective is a ritual. The feedback layer is a property of the system.

Feedback may arise:

• continuously,
• after a sprint,
• after an incident,
• after a customer interaction,
• after AI analysis,
• after a market change,
• after a values conflict,
• after detection of ecological or social impact.

An AIFC community must have a mechanism for:

• collecting signals,
• converting signals into change proposals,
• classifying proposals,
• determining the decision level,
• evaluating impact,
• approving or rejecting proposals,
• recording decisions,
• reflecting accepted changes in the Source of Truth.

Without a feedback layer, the community is rigid. Without a decision layer, feedback is chaotic.

AIFC requires both.

12. Interface layer

Communities exist alongside other communities.

Therefore, they need an interface.

A Community Interface describes:

• who the community is,
• what it offers,
• what it needs,
• which values it has,
• which boundaries it has,
• which inputs it accepts,
• which outputs it provides,
• how to cooperate with it,
• how conflict is escalated,
• how it accepts change proposals,
• how it announces the impacts of its decisions,
• how it protects sensitive information,
• how it allows AI agents to work within rules.

The interface layer enables cooperation between:

• member and team,
• team and department,
• department and company,
• company and customers,
• company and state,
• state and other states,
• human community and represented ecosystem.

AIFC does not describe isolated entities. It describes a network of communities with purpose.

13. Nested communities

An AIFC community may contain other communities.

A team may be a community. A department may be a community of communities. A company may be a community of departments. A state may be a community of municipalities, companies, institutions, and citizens. The world may be a community of states. Earth may be a community of human and non-human systems.

This model is recursive.

Each level may have:

• its own purpose,
• its own values,
• its own knowledge base,
• its own decisions,
• its own interface,
• its own AI governance,
• its own feedback loop.

At the same time, it may be part of a higher layer with shared values and rules.

14. Community levels

AIFC can be applied at different levels.

14.1 Team

A team has:

• concrete work,
• backlog,
• skills,
• retrospectives,
• AI capacity,
• local decisions,
• interface with other teams.

14.2 Enterprise

A company is a community of communities.

It contains, for example:

• marketing,
• sales,
• IT,
• security,
• finance,
• HR,
• customer service,
• product,
• operations.

Each unit may have its own knowledge base and its own Human Cockpit Layer over it.

The company as a whole shares values, strategy, governance, security rules, common interfaces, and a Company as a System model.

14.3 Country

A country is a wider community of communities.

It contains:

• citizens,
• municipalities,
• companies,
• schools,
• hospitals,
• authorities,
• institutions,
• professional groups,
• natural territories and resources represented through data or law.

AIFC does not mean a state governed by AI. It means a state that can structure knowledge, values, decisions, and feedback between communities.

14.4 World

The world is a community of states and global communities.

It contains:

• states,
• international organizations,
• scientific communities,
• global companies,
• cultural communities,
• humanitarian organizations.

At this level, AIFC describes interfaces, shared values, and feedback between communities, not centralized AI governance.

14.5 Earth

The Earth level also includes non-human systems:

• animals,
• plants,
• forests,
• rivers,
• soil,
• oceans,
• climate,
• future generations.

These systems may not have their own digital voice, but they can be represented through data, science, law, community representatives, or AI interpretation of signals.

AI may help translate signals from these systems into change proposals for human communities.

AIFC Earth does not mean a world governed by AI. It means a better ability for communities to perceive the impacts of their actions on the whole living system.

15. Role of AI in the community model

AI has several roles in an AIFC community.

15.1 AI as Accelerator

AI accelerates:

• understanding,
• synthesis,
• creation,
• analysis,
• documentation,
• solution design,
• knowledge work.

15.2 AI as signal detector

AI may detect:

• repeated problems,
• values conflicts,
• security risks,
• AI dependency,
• opportunities,
• weaknesses in the knowledge base,
• cross-community impacts,
• ecological or social signals.

15.3 AI as proposal generator

AI may propose:

• change proposals,
• workflow updates,
• skill updates,
• strategy alternatives,
• risk mitigations,
• non-AI workflow conversions,
• capacity allocation.

15.4 AI as External Expert Capacity

AI may act as external expert capacity if it has:

• purpose,
• scope,
• AI-NDA Boundary,
• cost control,
• human owner,
• audit,
• exit strategy.

15.5 AI as team member

An AI agent may act as a managed team member if it has:

• role,
• permissions,
• forbidden actions,
• approval rules,
• human owner,
• audit trail,
• ability to be turned off.

In all roles:

AI may propose. AI may accelerate. AI may warn. AI may help execute.

The community owns purpose and responsibility.

16. Minimal AIFC community model

A minimal AIFC community must have at least:

1. Purpose
2. Values
3. Human / community ownership
4. Source of Truth
5. Basic knowledge structure
6. Decision mechanism
7. Work structure
8. Feedback mechanism
9. AI usage rules
10. AI-NDA Boundary for non-public data
11. Fallback for critical AI workflows
12. Human Capability Reserve
13. Community Interface

This is the minimal model. An advanced community may add:

• AI capacity planning,
• AI operating modes,
• AI retrospective,
• Skill Evolution,
• workflow conversion,
• AI waste backlog,
• Company as a System model,
• multi-community governance,
• compliance levels,
• agent-actionable schemas and validators.

17. Anti-patterns

The AIFC community model rejects the following anti-patterns.

17.1 AI as owner of direction

AI proposes direction without clear human or community ownership.

17.2 Documentation without governance

The community has a lot of documentation, but lacks decisions, values, ownership, and feedback.

17.3 Top-down without feedback

Values and strategy flow downward, but experience and signals cannot change the system.

17.4 Feedback without decision

Everyone can propose changes, but there is no decision structure, so the system becomes overloaded.

17.5 AI-dependent operation

The community cannot perform critical or routine work without AI.

17.6 Knowledge trapped in tools

Know-how remains in chats, proprietary AI tools, personal accounts, or agent memories outside the Source of Truth.

17.7 Ghost AI community

The community has a digital facade and automated outputs, but lacks responsibility, values, fallback, and real ownership of purpose.

17.8 Community without interface

The community works in isolation and cannot clearly describe its boundaries, inputs, outputs, needs, and impacts toward other communities.

18. Summary

The AIFC community model describes a community as a living system.

Not as a tool. Not as an AI workflow. Not as documentation. Not as an organizational chart.

An AIFC community has:

Purpose
Values
Knowledge
Human Cockpit Layer
Decisions
Work
Learning
Feedback
Interface
Human Ownership
AI acceleration

The knowledge base is the memory and Operational DNA of the community.

The Human Cockpit Layer is human access to that memory.

AI is an accelerator over that memory.

The community remains the owner of purpose, values, decisions, and responsibility.

Its basic movement is twofold:

Top-down:
values -> purpose -> strategy -> work

Bottom-up:
experience -> signals -> change proposals -> decisions -> system updates

AI is not the ruler of the system.

AI is an accelerator, signal detector, proposal generator, external expert capacity, and sometimes a managed team member.

The community holds purpose, values, responsibility, and direction.

An AIFC community is purpose-driven, human-managed, feedback-enabled, and AI-accelerated.