§5. Scale Model
The three pillars (Memory, Access, Action) are not specific to individuals. They operate at every scale of cognitive organization. This is the fractal nature of Mind Extended: the same structural principles apply whether you’re looking at a single agent session or an entire industry.
The four scales
| Scale | What it is | Examples |
|---|---|---|
| Agent | An individual cognitive unit — human, artificial, or hybrid — that perceives, remembers, and acts | A Claude Code session. An operator. A skill. |
| Organization | A group of agents with shared purpose, shared memory, and coordinated action | A company. A personal system. A community. |
| Ecosystem | A network of organizations that interact, exchange value, and co-evolve | A holding company (services fund products, products demonstrate capabilities, personal system is the lab). A local tech scene. |
| Topology | The space of possibilities in which ecosystems exist — what can emerge, what is constrained | The current AI landscape (LLMs, MCP, agentic architectures). Market conditions. Cultural context. The extended mind thesis itself as a conceptual topology. |
Three pillars at every scale
| Memory | Access | Action | |
|---|---|---|---|
| Agent | Context window, auto-memory, session state | Tools, protocols, skills | Execute tasks, produce artifacts |
| Organization | Repos, protocols, brand guides, shared knowledge, institutional memory | Shared integrations, conventions, coordination channels | Project pipelines, business processes, service delivery |
| Ecosystem | Market knowledge, community norms, collective history, industry patterns | Public APIs, marketplaces, networks, communities, conferences | Market dynamics, community building, value exchange |
| Topology | Technology landscape, cultural context, economic conditions, physical laws | What connections are possible, what paths exist between states | What can emerge, what is constrained, what phase transitions are available |
Membership
Agents belong to one or more organizations. Organizations belong to one or more ecosystems. Ecosystems exist within a given topology.
The relationships are not strictly hierarchical — they’re overlapping:
- • An operator (H agent) may belong to multiple organizations simultaneously
- • An NH agent session belongs to whichever project it’s working on but carries cross-organizational memory
- • An organization can participate in multiple ecosystems at once
Consolidation at every scale
The feedback loop (Action → Registry → Memory → Planning → Action) operates at every scale:
- Agent: session activity → activity log → auto-memory → next session decisions
- Organization: project outcomes → institutional knowledge → strategic decisions → new projects
- Ecosystem: market signals → industry patterns → collective adaptation → new possibilities
- Topology: technological breakthroughs → landscape shifts → new topologies → new kinds of ecosystems become possible
The consolidation speed decreases as scale increases. An agent consolidates in minutes. An organization in weeks. An ecosystem in months or years. A topology shifts over years or decades.