ScribeShotOrganize a Knowledge Base with AI: Heptabase Workflow Notes
This real ScribeShot run captures a Heptabase workflow for letting an AI agent organize sources at scale. The source video stays beside the generated summary, timestamped chapters, AI notes and mind map.
Real ScribeShot run. This run produced a summary, timestamped notes and a mind map. No screenshots were captured, so the workspace reflects that rather than fabricating visual output.
Explore the generated ScribeShot workspace
Switch between Chapters, AI Notes and Mindmap. Timestamp buttons jump to the matching point in the original video.
The workflow addresses a scaling problem in personal knowledge management: manually filing hundreds or thousands of sources becomes expensive. An AI agent can learn organization rules, create cards from sources, place them into relevant whiteboards and refactor the structure as the collection grows.
- Initialize the knowledge base and define rules in an instructions card.
- Save web sources as web cards and YouTube sources as video cards.
- Use existing structure to infer where new cards belong.
- Place content onto topic-specific sub-whiteboards.
- Merge or split whiteboards over time as the knowledge base evolves.
- The goal is to organize sources, notes and the larger knowledge base with an AI agent.
- The bottleneck is scale: manual filing becomes hard when the collection reaches hundreds or thousands of items.
- Start by defining how the agent should organize information.
- The agent can inspect the existing knowledge base to learn current patterns and rules.
- The setup recommends top-level spaces such as Projects, Topics and Life.
- An instructions card acts as the rulebook for creating sub-whiteboards and placing new material.
- Links pasted into chat become saved cards.
- Web links become web cards; YouTube links become video cards with offline transcript access.
- Timestamps make it possible to locate specific points inside the saved video source.
- The AI places new cards on the relevant sub-whiteboard based on its learned organization rules.
- The example places an interview video on a computer-history whiteboard.
- The structure is not fixed forever.
- The agent can merge related whiteboards or split overly large ones as the knowledge base changes.
What this ScribeShot run demonstrates
Build your own reusable video knowledge base
ScribeShot keeps summaries, timestamped notes, mind maps and captured visuals beside the source video so useful ideas do not disappear after you watch them.
Source & attribution: These are editorially condensed notes generated from the linked Heptabase video. They do not republish the full transcript. Watch the original video for the complete product demonstration.