brain-ops

Ground agent answers by enforcing a brain-first read, enrich, and write loop that queries an internal context store before any external research. This workflow ensures every entity-centric response references existing knowledge rather than generating answers from scratch.

Ambient enrichment runs continuously in the background so inbound signals can spawn automatic entity detection, page loading, and graph reconciliation without interrupting the user. It enriches people, company, or topic data seamlessly while maintaining source attribution.

Prevent broken knowledge graph trails by enforcing mandatory back-linking, which ensures every mention of a person or company links back from the entity's page to the mentioning page. This maintains graph integrity during the auto-link reconciliation process.

Context grounding requires a read-before-external-call constraint to ensure the agent searches the brain context store for people, companies, or topics before performing external research. This deterministic behavior contract applies to inbound messages, meetings, and citations.

The best way to apply source attribution is through deterministic tool usage that writes new information back to the internal context store with proper citations. This ensures every enriched fact is tracked to its original inbound signal or external source.

Yes, entity detection and page loading run as part of an ambient enrichment loop that continuously processes inbound signals. This allows background graph reconciliation and timeline context updates without requiring manual intervention or interrupting user interactions.