The Orvessian platform
A governance workspace for the people responsible for AI activity, decisions and outcomes. Backed by evidence they can independently check.
From overview to evidence
Give operations a clearer view of activity, engineering a way to compare outcomes, and governance reviewers a path to the evidence.
Illustrative preview shown. The public demo uses synthetic data; authenticated evidence verification is in private testnet validation.
| Agent / deployment | Decisions | Latest reported action |
|---|---|---|
| Purchasing assistant procurement-v2 | 480 | Approval requested |
| Support triage support-v3 | 640 | Case routed |
| 10 other reporting agents | 160 | Decision recorded |
This example shows submitted context, not an independently detected policy breach. The customer's application enforces the limit.
Activity & decisions
Review agents reporting in the selected window, decision activity and model deployments. Recorded activity gives teams a common starting point; it is not automatic discovery of every running agent.
Outcomes & comparisons
Compare compatible labelled outcomes across deployment cohorts. Keep sample sizes, missing outcomes and evaluator definitions visible. A comparison shows a difference, not proof that a model change caused it.
Incidents & investigations
Review submitted incident and control observations alongside individual records. Source material stays in customer systems. Automated detection, notifications and incident assignment are planned additions.
Evidence & verification
In the authenticated pilot, open a record’s signed receipt, Merkle membership proof and Base verification state. Export record-specific evidence for independent checking without disclosing other records in its batch.
Agent governance, in practice
Bring the people responsible for AI around a shared record of activity, outcomes and exceptions. Build a governance process that keeps pace with the work.
Which agents are reporting, and what decisions are they recording? Start with the activity you receive and make gaps visible.
Pilot: scoped activity reportingFollow recorded actions, model versions, policy context and labelled outcomes. Investigate exceptions with the evidence in view.
Pilot: decisions, outcomes & submitted incidentsGive incidents an owner. Set tolerances, deliver alerts and retain a resolution history so reviews become repeatable work.
Planned: rules, alerts & incident workflowExport selected records with signatures and proofs. Check their integrity against a blockchain anchor outside the portal.
Pilot: independent evidence verificationOur roadmap adds agent owners, purpose and heartbeat status. Reporting coverage is not automatic discovery; immediate permissions and enforcement remain in your agent runtime.
The Orvessian difference
Cryptographic commitments on the Base blockchain create a public reference for private evidence. Authorised reviewers can detect changes to anchored records by checking the disclosed record and proof.
Inspect the public testnet anchors →Structured records, signatures and proof openings are held off chain under access controls. A batch commitment goes on chain, rather than your source conversations or documents.
The portal separates record integrity, recording-service signature, batch membership and chain confirmation. A valid anchor does not prove source truth, capture completeness or the safety of a decision. The entire application and database are not immutable.
Next on the roadmap
Agent ownership and heartbeats. Configurable thresholds and alert delivery. Incident acknowledgment and resolution. Saved views and scheduled reports. Easier integration with existing agent telemetry.
These are planned capabilities. Production identity, storage, backups and security validation remain release gates.
See proposed subscription plans →Common questions
No. The current workspace shows agents and events submitted to it within a selected window. It does not inventory every running system.
No. It supports an integrity and inclusion check for a disclosed record. The original source, completeness of capture and quality of the decision need separate review. See the verification boundaries.
A structured decision trail can support an organisation’s evidence process, subject to its own legal assessment and controls. Read our EU AI Act record-keeping guide for the relevant provisions and current timeline.