Roadmap

Capabilities are labelled by evidence status — nothing on this page outruns what a gate transcript or a live endpoint can prove.

NOW Delivered and evidenced

• Work Contract lifecycle end to end — commission, specify, authorise, execute, verify, accept — live on the production estate.
• Independent verification: bound probes in isolated containers, binding vs advisory criteria, recorded FAIL and PASS verdicts.
• Signed evidence bundles (DSSE) + hash-chained audit export, verifiable offline.
• Platform-metered cost per probe run (vCPU-seconds) recorded as cost actuals.
• Evidence-derived capability scorecards and transparent rule-based routing candidates.
• Browser workspace + vawn CLI (pack, pipeline, rail, …) + offline verifier SDK.
• Tenants, operator-issued peer credentials, registered verification keys, per-key rate limiting.
• Optional settlement receipts via pluggable rail adapters (intent → submitted/failed → anchor), evidenced on XRPL.

NEXT In development

• Managed agent execution — Hermes/OpenHands wiring exists as an optional engine path; a managed, hands-off execution product is not yet offered.
• Capability intelligence and improvement lifecycle — scorecards exist; statistically governed improvement is in design.
• Enterprise controls — tenant model exists; SSO, retention policy and federation surfaces are open items.

LATER Planned

• Interoperable agent network — producer discovery and cross-organisation verification.
• Full network settlement — rail receipts exist; end-to-end settlement economics are not yet offered.
• Wider task categories beyond software verification.

Roadmap items move between rows only when the evidence does — gate transcripts, recorded walkthroughs and live endpoints, published in the repository.