ENIGMA Axiom — Independent assurance for AI-native organizations & government

Independent AI assurance · Goals · Agents

Build the structure. Verify the agents.

ENIGMA Axiom is the independent assurance layer for AI-native organizations and government. We help you turn an objective into a verified structure of goals and tasks — then we independently check the AI agents acting within it, against authoritative sources and a panel of independent models. Never let an AI grade its own homework — or its own plan.

Free to try — no sign-up, 20 checks every 4 hours. Open-beta extension + MCP for agents.

One assurance layer, two jobs

AI now both sets the plan and executes it. ENIGMA Axiom keeps a human-trustworthy spine around both halves — so a machine never gets to be the only judge of its own work.

① Build the structure

Architect of verified goals

Turn an objective into a tree of goals, sub-goals and tasks. The E engine decomposes it and estimates whether the breakdown is sufficient to reach the goal, flags what's missing, and keeps a tamper-evident history of every change. Today the sufficiency check is an AI-assisted estimate that flags gaps. Formal, proof-backed sufficiency (rules + solver) is rolling out on supported, narrow domains.

② Verify the agents

Independent verifier of AI agents

Every claim, citation and action an agent produces is checked against authoritative databases — then cross-examined by a panel of independent models. Not the model that wrote it.

CrossrefPubMedOpenAlexarXivCourtListener

Independence is the whole point. An AI grading its own homework isn't assurance.

Live verifications

See the agent-verification layer

Paste an AI answer (or any claim with a DOI, arXiv ID or PubMed ID) and we'll tell you whether the sources are real and whether they say what was claimed.

How verification works

1

Extract

We pull every citation out of the answer — DOIs, arXiv IDs, PubMed IDs, case law.

2

Check it's real

Each citation is looked up in authoritative indexes — Crossref, PubMed, OpenAlex, arXiv, CourtListener.

3

Cross-examine

An independent panel of LLMs judges whether the claim matches the source — not the model that wrote it.

Flagship · Abu Dhabi · UAE

For the autonomous, AI-native government

Abu Dhabi is becoming the world's first AI-native government — where AI decomposes strategy into action and agents act with less human review at each step. ENIGMA Axiom is the independent layer that checks those decisions are sufficient, consistent and trustworthy before the state acts.

Get ENIGMA Axiom

The extension adds a verify badge and a goal panel to ChatGPT, Claude, Gemini and Deepseek — build goal trees, run the E decomposition, and verify any answer in place. Agents connect over MCP.

Chrome · open beta

For people — the browser extension

Chrome Web Store review is in progress. Join the open beta now via the source build, or email us for an early-access link.

Join the beta Source & install
API · MCP

For agents — verify over MCP / API

Give your own AI agents an independent verification and goal- sufficiency tool through the ENIGMA API and the Model Context Protocol.

Request API access API status

Pricing

Limits reset every 4 hours.

Free

$0

20 checks / 4h

  • No account needed
  • Citation verification
  • Goal builder + Chrome extension
Try now

Account

$0

30 checks / 4h

  • Everything in Free
  • History & saved goals
  • Higher limit
Sign up free

Teams, government or higher volume? Talk to us.