Agents

Finds semantic relationships between a link and your existing corpus.

Flags links that make conflicting or opposing claims within the same topic.

Scans the full corpus to surface under-explored topics and missing knowledge areas.

Reads signals from all agents and synthesises a ranked list of research clusters worth exploring.

Searches for new sources on a topic via Exa and ingests them into the corpus.

Re-scores unread links using cluster directives, active gaps, and decay status to surface what to read next.

Synthesises ideas from your top research cluster into a structured content draft — outline, prose, and key themes.

Produces structured feedback on a draft — strengths, weak claims, missing context, and a single recommended focus.

Rewrites a draft for a target publishing channel — LinkedIn post, newsletter section, or X/Twitter thread.

Scores knowledge gaps for commercial potential — surfaces which gaps represent topics you know deeply enough to turn into a course, guide, or consulting offer.

Runs four specialist Haiku critique lenses on a draft — argument, evidence, audience, and structure — then aggregates an overall score.

Reads unprocessed dev notes and extracts structured tasks — features, bugs, and improvements — via Claude Haiku.