Foundations

Ten philosophers, one comb.

EigenVertex doesn't run on a generic swarm of agents. It runs on ten regimes of intelligence — the modes of thought the Greeks named, distinguished, and never confused.

One mind lets things slip.
Ten disciplined minds form a comb — nothing that matters passes through unseen, unproven, unchallenged or untimed.

Aisthêsis · perceive
The idea

Why regimes, not roles.

Every platform in 2026 calls itself “multi-agent.” When everyone has agents, having agents means nothing. The useful question isn't “do we have agents?” — it's “what defines one, and why this cut?”

Most systems split their agents by business domain, or by pipeline step. Those cuts follow the org chart or the plumbing — not the nature of the thinking. EigenVertex cuts along the regimes of intelligence themselves: the fundamental modes by which a mind works a complex domain.

These regimes aren't a house invention. They were inventoried and distinguished by Greek thought, later restored by scholars like Detienne and Vernant on mètis. They are foundations — categories that survived twenty-five centuries because they cut cognitive work at its real joints.

Perceiving isn't remembering. Remembering isn't proving. Proving isn't reasoning. Reasoning isn't refuting. Refuting isn't finding the angle. Finding the angle isn't seizing the moment. Seizing the moment isn't judging. Judging isn't taking the long view. And none of that is yet producing. Ten distinct operations — and a system that blurs them does all ten poorly.

The ten

Each a tooth of the comb.

1Perceive & remember
Capture raw signals
Aisthêsis
Hold domain memory
Mnêmê
2Verify & reason
Qualify the evidence
Epistêmê
Structure the reasoning
Logos
3Challenge & angle
Test contradictions
Elenchos
Find the oblique angle
Mètis
4Decide & produce
Catch the right moment
Kairos
Weigh the decision
Phronêsis
Take the long view
Sophia
Produce the artefact
Technê
A clarification worth making

The philosophers are not the framework.

People sometimes hear “ten agents” and assume it's yet another agent framework. It isn't. The philosophers say what the system must be able to do. The frameworks say how we build it. We keep the two strictly apart — and we use the best standard tooling for the how.

The what · ours

Ten regimes of intelligence

A doctrine of separation. Which cognitive operations must never be confused — perceive, prove, refute, time, judge, produce. This is the part no framework ships.

AisthêsisMnêmêEpistêmêLogosElenchosMètisKairosPhronêsisSophiaTechnê
The how · standard

Proven agent frameworks

The plumbing is not where we try to be original. We build on the tools the industry already trusts — the ones IBM itself points to for production-grade, role-based orchestration.

LangGraphLlamaIndexLangChainCrewAISemantic Kernel

Frameworks orchestrate agents. EigenVertex orchestrates regimes of intelligence.

See the comb at work.

The ten regimes come alive around your projects and meetings — perceiving, proving, challenging, timing, producing.