This page explains how the research at Agentic Founder works. The short version: I build things, observe what happens, develop frameworks from those observations, and publish what I learn.

The Cycle

The research follows a consistent pattern:

Build — Implement an organizational structure, governance mechanism, or coordination pattern in OpenEnterprise. These are not thought experiments. They are working systems with real operational consequences.

Observe — Watch how agents behave within the structure. Note where things work, where they break, where unexpected patterns emerge. The most valuable insights usually come from failures.

Formalize — When a pattern recurs or a principle becomes clear, I develop it into a named framework. The Autonomy Chain, the Discipline Stack, the Heartbeat Organization, the Capability-Governance Gap — all of these emerged from repeated observation, not from theoretical reasoning.

Publish — Document the framework on Agentic Founder. This serves two purposes: it forces clarity of thought, and it opens the work to outside scrutiny. Both make the ideas stronger.

Iterate — Publication often reveals gaps in the framework. Reader questions, my own dissatisfaction with the writing, or new experiments in OpenEnterprise feed back into the cycle.

What Gets Published Where

The type of publication depends on the maturity of the idea:

• Notes — early observations, partial thoughts, questions without answers yet
• Discoveries — patterns or findings with enough substance to document but not enough to claim as settled
• Essays — developed arguments that make a case for a specific position
• Theses — foundational frameworks intended as reference points for the broader body of work
• Lab entries — methodology, environment descriptions, and experiment documentation

Research Artifacts

Some frameworks developed at Agentic Founder may also appear as open source reference implementations. When they do, they'll be linked here.

Active and planned repositories:

• agentic-autonomy-chain — reference implementation of the full decision loop
• discipline-stack — governance layer components
• heartbeat-organization — temporal coordination system
• append-only-learning — operational knowledge accumulation

These are research artifacts, not production software. They exist to make the ideas concrete and testable.

The Environment

OpenEnterprise is the primary testing ground. It provides a real operational context where agents hold genuine authority, make consequential decisions, and operate under the governance frameworks being researched.

The environment is not static. It evolves as the research progresses. What I publish on Agentic Founder reflects the current state of understanding, which is always provisional.