Axiom Workbench The Evidence Engine for Methods

DETERMINISTIC · SEALED PACKS · SHA-256 HASHES

Run a method. Keep the record.

Drop in a method and a dataset. Axiom runs it against causal baselines on the same data under the same rules, then seals a pack — summary, metrics, plots, and SHA‑256 content hashes.

If evidence is incomplete, the run is refused and the gaps are listed. No silent fallback. Not AI.

Run a mission → What is Axiom?

API key required · Metrics from the method run only · UI never re-scores · Not AI

Mission

Live domains available now. More industries in staging.

Thresholds (into method env; UI never re-scores)
Options

Deliverable is the sealed pack (numbers, plots, plain‑English summary, hash‑sealed inventory, zip). Local default payment provider is mock — no Stripe account required.

Status

  1. Idle
  2. Running
  3. Pack ready
  4. Failed

Idle — ready for a mission.

FAQ

Do I need an API key?

Yes in production. Paste your key in the mission form. Without a valid key the evaluate path returns unauthorized — full packs are not free-open.

What is in a sealed pack?

Summary, metrics, plots when present, and a hash‑sealed inventory (SHA‑256 of each artefact). Recompute digests to verify files still match the manifest. The UI never re-scores.

Why was my run refused?

Incomplete evidence, policy failure, or missing required series/baselines. Gaps are listed when the engine refuses — no silent fallback. Not AI guessing.

Which domains can I run?

Live today: industrial actuators/SCADA, robotics (flexible joints), financial telemetry, and power electronics / motor drives. More industries remain in staging (Coming soon in the form).

Is this method-agnostic?

Yes. Candidate_Zero / Reference_Standard are evaluation labels, not the product brand. Full product story: Labs Axiom page.