cold-start

Bootstrapping a blank knowledge graph requires sequencing high-leverage data imports to populate core data quickly. This workflow applies phased ingestion to Markdown/Obsidian sources, contacts, calendars, Gmail, and social archives, linking entities to create a connected base.

Importing Markdown and Obsidian sources is best handled through a phased data ingestion workflow. This process creates pages for people, companies, and meetings while propagating entity links and timelines across phases to build a connected knowledge base.

Entity linking during contacts and calendar imports works by propagating connections across ingestion phases. The workflow tracks progress, creates dedicated pages for people and meetings, and automatically links these entities to build a structured, searchable knowledge graph.

You can import Gmail threads and X/Twitter archives without exposing raw credentials by using a credential gateway. A safety module vaults credentials, enforces scope, logs actions, and requires human approval for destructive operations during ingestion.

Safety gates during phased data ingestion enforce consent-driven steps by vaulting credentials, limiting scope, and logging actions. The workflow requires explicit human approval for destructive operations, ensuring sensitive data imports remain secure and controlled.

Consent-driven data import works with chat exports and social archives by gating sensitive ingestion steps behind a credential gateway. This phased approach tracks progress and links entities across sources while maintaining strict safety and scope enforcement.