Structured Reports

Beyond answering ad-hoc questions, Sera can assemble structured reports on demand. Each report is backed by a carefully built question that runs across your full institutional memory, so you get a consistent, comprehensive readout rather than a one-off answer.

How to request one

Reports are generated through the POST /report endpoint, or through the dashboard proxy that calls it for you. You name the report type you want, and Sera runs the matching pre-built question and returns the result.

The six report types

• project_status — where initiatives stand, what is moving, and what is stalled.
• governance_audit — the state of circles, roles, decisions, and governance hygiene.
• member_activity — who has been active across recent memory.
• risk_summary — the open and emerging risks captured in memory.
• team_overview — a broad picture of the team and its structure.
• meeting_prep — a briefing to walk into a meeting well informed.

Saving a report to the Knowledge Base

Every report can optionally be preserved. Set the saveToKb flag and Sera creates a Draft KB article titled "Report: [Type] - [Date]". That draft then appears in the Memory Review Queue, where a human can review it and promote it into the Knowledge Base. This is how a useful report becomes durable memory rather than a throwaway view.

Security of the request

When you trigger a report from the dashboard, the API secret is injected server-side. It never travels to or lives in the browser, so generating reports from the dashboard does not expose the credential to anyone using it.

What reports can and cannot do

Structured reports are powerful summaries, not oracles. Keep three limits in mind:

1. They draw only on reviewed and processed data. A report reflects what has already been extracted and reviewed into memory. Material still sitting unprocessed, or drafts nobody has confirmed, will not appear.
2. They do not forecast. A report tells you the current and historical state of your memory. It does not predict outcomes or project the future.
3. They do not replace human judgment. A report organizes what is known; deciding what it means and what to do is still a human act.

Because of the first point, report quality tracks how completely the review queues have been worked. A circle that keeps the Memory Review Queue and its Candidate and Draft records current will get rich, trustworthy reports. A circle with a large unreviewed backlog will get reports that quietly understate reality. The most reliable way to improve your reports is to keep reviewing memory. Run reports often, but read them as a well-organized reflection of what you have captured, not as the final word.

Key points

Structured reports are generated on demand via POST /report or the dashboard proxy, each backed by a pre-built question over your full memory. The six types are project_status, governance_audit, member_activity, risk_summary, team_overview, and meeting_prep. An optional saveToKb flag creates a Draft KB article titled "Report: [Type] - [Date]" that then appears in the Memory Review Queue for promotion. The dashboard injects the API secret server-side, so no secret reaches the browser. Reports draw only on reviewed and processed data, do not forecast, and do not replace human judgment; their quality tracks how completely the team has reviewed its queues.