AI review
Understand where work stands, check the evidence, and turn the next decision into action.
AI review helps people and AI understand where work stands, what needs attention, and what to do next. Choose a case, personal, or team scope to bring the relevant situation, evidence, and proposed actions into one report.
You can also use the AI review API on its own: send data from your application and receive a review without creating cases or playbooks.
The value carries across evaluations. Record a result, reevaluate, and see whether the earlier conclusion still holds. For case evaluations, the next report considers the previous report, handled proposals, your feedback, and the current evidence.
Start with a case
Open a case in Web and select Evaluate in AI review. Case editors can request an evaluation; current case readers, including public readers, can read the saved report. Generation runs in the background. Reading a report never runs AI.
An evaluation can consider:
- The case’s purpose, input, and current situation.
- Tasks, their instructions and recorded outcomes.
- Non-deleted records and their structured data.
- The pinned playbook’s instructions and expected outputs, when readable by the case audience.
- Available saved text from linked external sources, including sources on public cases.
These inputs are bounded. A linked URL or attachment does not mean Review opened it, inspected an image, or executed a test. Missing and unavailable evidence remains unverified.
Check the work against its criteria
Review reviews the supplied results and records against the case’s purpose and, when available, the pinned playbook’s instructions and expected outputs. It looks for supporting evidence, gaps, and contradictions that matter to the decision. A missing cost estimate, for example, leaves a recommendation to accept a customer request unverified; a recorded estimate that exceeds the expected revenue provides evidence for reconsidering it.
The report distinguishes requirements stated explicitly in the case or playbook (stated) from criteria inferred by AI (inferred). Each checkable condition evaluates to:
| Verification status | Meaning |
|---|---|
passed |
Positive evidence supports the condition. |
failed |
Contradicted or unmet in recorded evidence. |
unknown |
Missing or incomplete evidence prevents a reliable check. |
Each verification card includes a concise explanation of what holds or what gap remains, alongside direct citations to supporting records, tasks, or playbook instructions. Inspect the cited evidence and any stated limitations before relying on the conclusion.
Read the report and its evidence
A report puts important information first: the current situation, meaningful changes, open issues, and useful next decisions. Its summary also appears in work lists.
The assessment distinguishes:
| Health | Meaning |
|---|---|
healthy |
Positive evidence shows meaningful progress or important conditions being satisfied. |
at_risk |
A specific emerging obstacle or deviation threatens the stated purpose. |
critical |
A confirmed material problem or blocker is actually preventing important work. |
unknown |
The supplied evidence cannot support a reliable diagnosis. |
health replaces the former verdict field in reports and history summaries. The summary explains the current situation and the main evidence-based reason for the diagnosis, in at most 280 characters. For unknown, it explains the evidence gap. Pending tasks, optional missing information or a lack of recent updates alone do not lower health; non-unknown diagnoses need supporting citations. Coverage limits matter when interpreting an overall diagnosis.
Follow cited records, tasks, saved source text, or the pinned playbook to inspect the basis for a claim. An AI judgment supports a decision; it does not replace a designated reviewer’s formal approval.
Turn a proposal into work
A report may propose four kinds of work. Review the full saved contents before confirming; creation uses them unchanged and accepts no title, content, or definition overrides. Creation is an explicit user or agent action; evaluation alone creates no work.
- Create a task to gather evidence, resolve a problem, or advance the next decision.
- Create a playbook suggestion to propose a repeatable improvement. The normal suggestion workflow decides whether it becomes a new playbook version.
- Start a case for a separate purpose, responsibility, or completion criterion. It creates a directed handoff from the reviewed case without closing it.
- Create a playbook with its first version for an evidenced recurring pattern and likely future reuse. A completed case or theoretical reuse alone is insufficient; handled Playbook-creation recommendations are not proposed again for the same case.
- Dismiss a proposal when it does not fit. Optional feedback informs the next evaluation.
Suggestion text is shared with the playbook audience, which may differ from the case audience. Review it for private case details before submitting it.
New cases and playbooks use your active billing home with restricted creator access. Source sharing and external sources are not copied. Confirm using only reportId, proposalId, and action; optional feedback is for dismissal. Retries use the proposal identity, so do not supply idempotencyKey.
Record the result, then reevaluate
For example, a customer request has clear commercial interest but no delivery or support cost estimate. Review points out the missing evidence. The team creates a task, checks the costs, and saves its findings as a record. A new evaluation can then reconsider whether the request makes sense and propose a useful next step.
A closed task is a signal, not proof that the underlying issue is resolved. Keep the actual result and evidence in the case. Handled proposals and feedback help later evaluations avoid repeating work already adopted or dismissed.
A freshness notice appears when meaningful evidence has changed since evaluation. It does not run AI automatically. Select Reevaluate to update the assessment. A notice is advisory: inspect the saved report and the newer evidence before acting.
Control, privacy, and credits
Evaluations currently run on demand.
The case owner’s AI settings control built-in AI permission. Custom home guidance is not supported; API, CLI and MCP generation can supply optional evaluation criteria. Generation uses token-priced credits from that owning account and charges only when a report is saved. Task, suggestion, or playbook creation costs 5 credits. A new case costs 5 credits plus 2 for its handoff. These actions charge the acting caller’s active billing home; dismissal is free.
Current case readers, including public readers, can see saved reports and linked sources, including content created before a sharing change. Ownership transfer clears derived reports and action links. Durable conclusions that must outlast a report belong in records.
Use Web, CLI, or MCP
The same saved report and proposal decisions are available across surfaces.
epismo report case generate "$CASE_ID"
epismo report case get "$CASE_ID"
epismo report case act "$CASE_ID" \
--report-id "$REPORT_ID" --proposal-id "$PROPOSAL_ID" \
--action createGeneration returns job metadata. Poll get until job.status is complete, failed, or skipped. For MCP use epismo_report_case_generate, epismo_report_case_get, and epismo_report_case_act.
Browse saved evaluations
In a case's AI review in Web, select History to browse saved dates, health diagnoses, and summaries, then open a full report with its evidence and proposals. History views do not invoke AI. Case detail includes the latest readable report and generation job; older evaluations are fetched only when requested.
History reads require current case access, including public access, and include reports for the current owning account. Superseded reports have historical: true and are read-only: creating or dismissing an unhandled proposal from one returns HTTP 409. Read the latest report and reconsider its proposal before acting. Changes to evidence alone still show an advisory reevaluation notice.
epismo report case list CASE_ID --page-size 20
epismo report case list CASE_ID --cursor NEXT_CURSOR
epismo report case get CASE_ID --report-id REPORT_IDPersonal and team reports
Personal and team reports are available through CLI, API and MCP for generation, reads, history and proposal handling. Web supports case reports, but has no personal or team report UI.
Personal reports
Personal review assesses your assigned open cases, assigned open tasks in open cases, and open suggestions in your managed playbook inbox. Parent cases may be assigned to someone else but must be open. Only work owned by the active workspace, or your own account in personal space, is included. Sharing from another space and public-only access do not qualify. Home team filters, the active tab and pagination do not narrow the review.
Collection is bounded and includes supporting tasks and recent records. It does not inspect external sources or playbook definitions. Only the generating user can read the saved personal report in its original billing home, even if source sharing or AI policy later changes.
Team reports
Team review assesses open cases explicitly shared with the selected team and owned by the active workspace, their tasks and recent records, and open suggestions on team-shared playbooks owned by that workspace. Work shared only with a person or another team, and work owned by another space, is excluded. Collection is bounded and does not inspect external sources or playbook definitions.
Current team membership in the active workspace is required to generate or read a report, browse its history, or handle a proposal. Reports and handling decisions are shared with current members, regardless of who generated them. Generation uses workspace credits and the data owner's built-in AI policy; membership, source sharing and policy are rechecked before publication.
Saved team text remains readable by current members after source sharing, archival or deletion changes. Evidence labels and links still require live source access. Consider that durable audience when generating a team report.
Personal and team proposal handling
Personal and team generation assess current work and consider feedback on earlier reports. Previous report contents, proposal handling history and dismissal feedback are not supplied as evidence. A new evaluation assigns new proposal identities and may recommend similar work again. Saved history remains readable without a 14-day cutoff; check existing work before confirming another recommendation. Case evaluations, described above, can use prior reports and handling history.
Both scopes return at most three proposals and never execute them automatically. Confirm the full saved contents unchanged or dismiss with optional feedback. Creation checks current destination permissions. Tasks target the specified case and suggestions target the specified playbook/version. New cases are ad hoc without an implicit handoff; new cases and playbooks retain creator-private access without copying source sharing. A newer report makes older unhandled proposals read-only.
Team creation, its charge and the shared handling decision commit together, preventing duplicate work when another member retries. Generation returns a job; reuse the same generation idempotency key after an uncertain response, and poll until it completes, fails or is skipped. Reading saved reports never runs a model.
Reference
- AI settings: permissions and settings.
- API conventions: report endpoints and request fields.
- CLI: commands, arguments and retries.
- MCP: report tools and their arguments.
Case, personal and team report generation accepts optional criteria (CLI: --criteria). Delete a report with DELETE /v1/reports/{reportId}, epismo report delete REPORT_ID, or epismo_report_delete. Case editors, personal report owners, and current team members respectively may delete. Content is erased; history retains the deletion time, while decisions and created work remain.