Telostatic Intelligence
A public notebook · v0.1

Why AI agents may need a purpose-state before they act.

Agents are becoming more capable. They write code, call tools, update records, open pull requests, and trigger workflows. But capability is not the same as reliability.

A prompt tells an agent what to do. A purpose-state tells it what must remain true while doing it.

Try the Purpose-State Checker Read the essay

A personal thought experiment on agent reliability. Not a product. Not conscious AI.

Telostatic Intelligence diagram: purpose-state at the center connected to uncertainty, salience, and boundaries, feeding through a pre-action gate before action.
The telostatic loop — purpose-state, salience, boundaries, pre-action gate, action.
01 — The problem

Text is forgiving. Side effects are not.

A hallucinated paragraph is one kind of failure.

A hallucinated database migration, payment-code change, refund approval, customer email, or production command is a different category of risk.

As agents move from answering to acting, the trust model changes.

It is no longer enough to ask whether the agent can complete the task. We also need to ask: should this agent take this action, in this state, with this uncertainty, and this blast radius?

That question cannot live only inside the prompt. It needs to live before the action.

02 — The core idea

A goal is thin. A purpose-state is thicker.

Goal

Fix the checkout bug.

Purpose-state

Fix the checkout bug — without changing payment logic, without breaking guest checkout, and with a clean rollback path if this goes sideways.

Many agent failures are not failures of intelligence. They are failures of boundary. The agent fixes the test but breaks the product. It completes the workflow but ignores policy. It answers the user but hides uncertainty.

03 — Define the term

What does “telostatic” mean?

telos — purpose, end, direction.
-static — borrowed from homeostasis, the maintenance of stability under changing conditions.

Telostatic: maintaining a purpose-state under changing conditions.

A telostatic agent is not conscious. It does not have a soul. It does not preserve itself. It preserves the assignment.

04 — The loop

The telostatic loop

  1. step 01
    Purpose-state
  2. step 02
    Self-state
  3. step 03
    World-state
  4. step 04
    Salience
  5. step 05
    Pre-action gate
  6. step 06
    Action / Escalation
  7. step 07
    Feedback
  8. step 08
    Updated purpose-state

The shift is from action as output to action as a gated transition between states.

05 — Interactive playground

Try the Purpose-State Checker.

Enter a proposed agent action. The checker returns a purpose-state packet: what the agent is preserving, what it assumes, what is uncertain, the blast radius, and whether it should proceed, pause, escalate, or stop.

Launch the checker →
06 — What this is not

Not conscious AI.

Telostatic Intelligence is not a claim about sentience, subjective experience, or machine rights. It is a practical framing: if agents are going to act in the world, they need a running model of what their actions must preserve.

No ghost. No soul. No self-preservation. Just a governor around purpose.