Infoconex AI Flywheel Specification v0.2.1

Principle Research Dossiers

This section does not define the AI Flywheel principles. The official principles are maintained in the Infoconex AI Flywheel Specification.

These research pages investigate the ideas behind each principle. Their purpose is to record:

  • related concepts and prior art,
  • historical origins and dates that still need to be verified,
  • evidence gathered from related frameworks,
  • similarities and differences that may affect claims about what is different,
  • unanswered questions,
  • evidence that still needs to be collected,
  • and the current research position based on what is known so far.

A research page may remain incomplete or change as new evidence is found. The specification remains the source for what the AI Flywheel is.

Table of Contents

  1. Autonomy Is Bounded by Human Authority — Investigates authority given to AI, supervisory control, approval gates, uncertainty escalation, authority boundaries, and governance of self-improving systems.
  2. AI Is the Operator, Not Merely the Assistant — Investigates the shift from AI-generated recommendations to AI directly operating persistent processes and capabilities.
  3. Work Is Distributed Across a Moving Determinism Boundary — Investigates prior art for moving responsibility among reasoning, procedure, and deterministic execution.
  4. The SOP Is an Operational Control Plane — Investigates machine-consumable procedures, runbooks, skills, plans, and workflow definitions as persistent controls for operation.
  5. Execution Must Produce Outcome Evidence — Investigates observability, external validation, feedback signals, execution traces, and evidence needed to judge real outcomes.
  6. Failure Determines Where the System Evolves — Investigates whether existing systems identify weaknesses and send improvements to different persistent destinations.
  7. Learning Must Change a Persistent Operational Asset — Investigates how systems turn execution experience into durable assets that future operation can use.
  8. Improvement Must Compound Through Reuse — Investigates how repeated execution builds reusable capability and whether those improvements can span multiple operating mechanisms.

Research Status Language

The research pages use four broad states:

  • Established: Supported by clear primary-source evidence.
  • Current evidence: Supported by the framework and source review completed so far.
  • Open: A research question that has not been answered well enough yet.
  • Not established: The reviewed sources do not currently demonstrate the mechanism; this does not prove that no implementation exists.

For framework-by-framework evidence, see the framework and prior-art research section and framework comparison matrix.