Knowledge Graph Direction
ThinkWork’s current Memory story does not depend on a full knowledge graph product surface.
Today, the shipped story is simpler:
- threads are the system of record
- document knowledge handles retrieved source material
- long-term memory carries learned context forward
- retrieval and context assembly decide what reaches a turn
A knowledge graph is the likely next step beyond that, not the current center of the product.
Why this direction matters
Section titled “Why this direction matters”A graph-shaped memory layer could make recall more relationship-aware by modeling entities and connections directly, such as:
- customers and accounts
- services and dependencies
- incidents and linked decisions
- people, teams, and ownership
- concepts and their relationships over time
That could make memory more structured than document chunk retrieval or flat fact recall alone.
How to think about it today
Section titled “How to think about it today”Treat this as forward direction, not as something the current docs ask you to depend on.
If you are evaluating ThinkWork now, plan around:
- Threads for the work record
- Memory for context
- Document knowledge for source material
- Long-term memory for learned carry-forward context
Then watch the roadmap for richer structured-memory and graph capabilities.