packages/db/src/seedDemo.ts.Observe → update memory → plan → execute (propose-only) → measure and learn — on a schedule, with no manual prompting required. Never sends outreach or publishes externally; every draft still lands in an approval queue for a human.
Autonomous loop disabled (kill switch)
The scheduler will not start a cycle automatically. Manual "Run now" still works.
Next scheduled run
Scheduler is off (kill switch).
Awaiting human approval
Researched prospects
0
Opportunities
0
Outreach/content drafts
0
No cycle has run yet
Enable the loop above, or click Run now to trigger the first observe -> plan -> execute -> measure pass manually.
The knowledge graph is the company's reasoning memory — confidence here reflects real observed evidence (replies, deliverability, activation outcomes, shipped fixes), not just editorial approvals. The Executive Planner increasingly reasons from this, not from each package's own operational tables.
Confidence by entity type
| Entity type | Nodes | Evidenced | Avg confidence (evidenced) |
|---|---|---|---|
| Companies | 0 | 0 | — |
| Prospects | 0 | 0 | — |
| Buyer contacts | 0 | 0 | — |
| Growth assets | 0 | 0 | — |
| Growth opportunities | 0 | 0 | — |
| Hypotheses | 0 | 0 | — |
| Email messages | 0 | 0 | — |
| Search queries | 0 | 0 | — |
Hypothesis confirmation
Confirmed
0
Refuted
0
Inconclusive
0
Unresolved
0
Confirmation rate
Not enough measured hypotheses yet
Buyer role performance
Which buyer characteristics correlate with a positive outcome.
No evidenced buyer contacts yet.
Needs attention
Evidenced nodes with real contradicting signal, lowest confidence first.
Nothing evidenced and low-confidence right now.