Skip to contentSkip to main content
OfferAgent
Open menu

Evidence registry

Verify what OfferAgent does in production

This registry connects verified product claims to the public product page, responsible technical contracts, reviewer and current review period. It contains no outcome claim without an approved dataset.

Active evidence objects: 5Reviewed: Next review:

Verified product claims

Each claim is blocked from a new release when its review expires or either localized product page is no longer published.

capability.form-engine.versioned-intake.v1

Published forms validate structured answers and retain the form version and submission source.

Active and reviewed

Explicit boundary

A form does not create or send a quote unless a separately configured workflow permits it.

Review public product descriptionOpen signed-in product view
Reviewer
OfferAgent Product and Evidence Review
Reviewed / Next review
/
Permanent public source

Responsible contracts

Frontend
form-engine.public-runtime.v1
Backend
form-submission.canonical-input.v1
Database
form-engine.versioned-definition.v1

Measured outcome: No result published.

capability.quote.immutable-revisions.v2

Sent and accepted quote revisions remain traceable while changed terms are saved as a new revision.

Active and reviewed

Explicit boundary

A draft can be edited, but a published customer version is not silently overwritten.

Review public product descriptionOpen signed-in product view
Reviewer
OfferAgent Product and Evidence Review
Reviewed / Next review
/
Permanent public source

Responsible contracts

Frontend
quote-revision-workspace.v2
Backend
quote-revision.lifecycle.v2
Database
quote-revision.immutability.v1

Measured outcome: No result published.

capability.work-queue.next-action.v1

The seller work queue combines customer activity, ownership and timing into a visible next action.

Active and reviewed

Explicit boundary

The suggested next action depends on available events and configuration; it is not a guarantee of conversion.

Review public product descriptionOpen signed-in product view
Reviewer
OfferAgent Product and Evidence Review
Reviewed / Next review
/
Permanent public source

Responsible contracts

Frontend
seller-work-queue.v1
Backend
work-queue.projection.v1
Database
event-log.canonical.v1

Measured outcome: No result published.

capability.automation.controlled-sequences.v1

Published automation sequences expose lifecycle, next run and stop reasons and can stop on customer reply or acceptance.

Active and reviewed

Explicit boundary

A sequence runs only after configuration and publication; the platform does not imply delivery when a provider rejects a message.

Review public product descriptionOpen signed-in product view
Reviewer
OfferAgent Product and Evidence Review
Reviewed / Next review
/
Permanent public source

Responsible contracts

Frontend
automation-sequence.workspace.v1
Backend
automation-sequence.execution.v1
Database
automation-sequence.outbox.v1

Measured outcome: No result published.

capability.ai.advisory-grounding.v1

AI recommendations retain evidence references and remain advisory until a user explicitly chooses an action.

Active and reviewed

Explicit boundary

AI does not independently set final prices, send quotes or execute recommendations.

Review public product descriptionOpen signed-in product view
Reviewer
OfferAgent Product and Evidence Review
Reviewed / Next review
/
Permanent public source

Responsible contracts

Frontend
ai-recommendation.review.v1
Backend
ai-recommendation.grounded-output.v1
Database
ai-recommendation.audit.v1

Measured outcome: No result published.

Outcomes require a separate research gate

Product evidence shows that a capability exists and where its boundary lies. It does not prove a customer outcome. Benchmark results publish only after consented aggregation, minimum sample gates and method review.

Read the methodology