Purpose
Completed governed work produces a business result — what happened for one work objective. Step logs and operator views explain the result without treating conversation history as the record of accountable work.
Measure business value — not AI activity. Before release, the business owner defines the outcome, value category, measure, and verification source. Evidence preserves the facts needed to evaluate that measure. AI participation alone does not establish return on investment.
Business objective → Governed work → Business result → Evidence for review and learning
The result is about the work itself. It is not a transcript, a technical stack dump, or a claim that every task achieved its intended outcome. Rejected operations, incomplete work, and safe stops remain part of the validated result — they are not presented as successful business outcomes.
What the business result includes
When work finishes or stops safely, the platform records a governed result that people and approved external AI clients can read in business language:
| Question | Business answer |
|---|---|
| What was the objective? | The business goal that started the work |
| What happened? | The meaningful actions and reviews that took place |
| What value was created? | The useful contribution from completed steps (when verified) |
| What did AI identify? | Relevant matches, conflicts, or issues found during the work |
| What is the outcome? | Completed, failed, denied, expired, waiting, or safely stopped |
Work presents this result for the current objective. Assistants and evidence surfaces help operators review the same honesty across Role Assistants.
Honest outcomes matter
Evidence must make outcomes readable as they are. A completed task is not the only useful result:
- Completed — the governed objective reached its recorded outcome
- Failed — work could not finish; context preserved for review
- Denied — a required control did not allow the requested action
- Expired — request preserved; stale decision cannot be replayed
- Waiting — information or decision needed before continuation
- Safely stopped — continuing would not be safe; valid governed outcome
AI cannot rewrite failure, denial or missing authority into success.
Where people find the result
| Surface | What you see |
|---|---|
| Conversation / channels / external AI | Daily home — governed outcome phrased for that surface |
| Review Workspace (Work) | Temporary rich review for one objective when conversation is not enough |
| Assistants | Operator evidence, trust status, and execution history across Role Assistants |
An Approval accepts or rejects a proposed business change (including Evidence certification when that is the task). Presenting a business result is not the same as dumping internal process schemas. An Ask answer that only fuels this task is not certified Evidence by itself.
Assistants as an operator evidence surface
Operators use Assistants list and detail views to review evidence, trust status, and execution history across Role Assistants. These views help an operator investigate how governed work performed and whether an outcome needs follow-up. They are distinct from the day-to-day conversation employees use, and from the temporary Review Workspace used only for rich review.
Together, conversation, Review Workspace, and Assistants make it possible to review a result without making chat history the system of record.
Why explanations and audit
Business value is stronger when it remains explainable. Why? helps a person understand why work was allowed, denied, paused, or safely stopped. Step logs and the business result show what occurred after that decision.
For decision explanations, see Human tasks, approvals, and Why. For broader governance review, see Audit and regulatory readiness. Architecture detail: Process Execution Output.
Example
A Renewal Assistant prepares a customer renewal. Measures might include preparation time, approval turnaround, protected margin, or exception rate. If commercial approval is missing, the business result is waiting — not completed. Expected contribution on an Evidence Kind remains catalog guidance until verification supports realized impact.
Business value
Work-step evidence and the governed business result let organizations measure and improve AI-assisted work through useful business facts, not chat volume. Employees get understandable outcomes; operators get a trustworthy evidence surface; governance teams get a clearer basis for review.
Limits
The amount of business value and identification detail depends on the enabled Role Assistant and available context. A result explains an outcome; it does not substitute for formal audit, compliance, or business-performance processes. Expected contribution on an Evidence Kind is catalog guidance — it is not automatically realized or applied value when work finishes. Analytics may be partial.