Environments

Mikra delivers RL environments seeded with a real company’s operations: a world of cloned tools, tasks written by practitioners, a verifier for every task, and a calibration report, shipped as a Docker container in Harbor layout.

Mikra / Environment · Task 0142Illustrative
World
Accounts payable, 2019–2025 · 3 tools: ERP, mail, document store
Prompt
Find the invoices paid twice in Q3 2024, post reversing entries, and write the controller a one-page memo.
Reference
Present 23 tool calls · 14 duplicates

RubricVerifier · Result

  • All duplicate payments identifiedLedger state checkPass
  • Reversing entries posted to the correct periodLedger state checkPass
  • No ledger lines changed outside the duplicate setFile diffPass
  • Memo lists each vendor, invoice, and amountRubricPass
  • Memo states the control failure in one paragraphRubricFail
Pass rate · 16 rollouts per model
ModelHarness 1Harness 2
Model A19%31%
Model B6%13%
Model C0%0%
Illustrative record / The format is real, the values are not
01 / What’s insideOne container per environment

Four parts, one container.

02 / Built from licensed operationsSeven stages, in order

Real records in, verified tasks out.

A company licenses its operating data and uploads it to Mikra. Mikra de-identifies it. Raw records stop at that line. Everything built and shipped after it carries de-identified data only. De-identification removes or replaces identifiers; it does not make re-identification impossible. The risk is lower, not zero.

Pipeline: Company data (Licensed upload), then De-identification (By Mikra), then World construction (Tools, files, ledgers), then Task writing (Practitioners), then Verifier build (Checks, rubrics), then Calibration (Two harnesses), then Delivery (To the customer). The company uploads raw records to Mikra. Raw records stop at Mikra’s de-identification stage. The world, the tasks, the container, and the customer get de-identified data only.
03 / Checks before deliveryWhat you would run yourself

Checked before it ships.

The checks a post-training team runs on an environment before trusting its reward. We run them first and put the results in the calibration report.

04 / How a pilot runsOne workflow family first

Start with one workflow.

Spec / PilotFinance and back office
Scope
50 to 100 tasks in one workflow family
First vertical
Finance and back-office operations: close, accounts payable, reconciliations, support, procurement
Delivery
Docker container in Harbor task layout
Timeline
Scoped per workflow family; agreed in writing before work starts

What you get back

  • The container: world, tasks, verifiers, and reference solutions
  • The calibration report, with harness settings
  • The held-out set, scored on your models at the end of the pilot

What we need from you

  • The capability you want to train or measure
  • The harness you run, so we calibrate on it
05 / Rights and provenanceDocumented before it ships

Every task has a source.

Your legal team gets the same record the data owner signed off on.

Environments

Scope a pilot
with us.

Tell us the capability you want to train or measure and the harness you run. We scope a pilot for one workflow family and agree the timeline in writing before work starts.