A personal knowledge graph stores the relationships between saved details, so an agent can explain why a detail matters.
Most personal data is stored as separate records: one note, one email, one meeting transcript, one screenshot. The useful context is often the relationship between those records. Maya is tied to the product launch. The product launch is tied to a pricing decision. The pricing decision is tied to a constraint from legal. The follow-up is tied to your next call.
What The Graph Holds
In graph language, a person, project, decision, promise, idea, or preference is an entity. A connection between them is an edge. The edge is where memory becomes useful: Maya owns the review, the review blocks the launch, the launch depends on the decision, and the decision came from a specific brief.
This is why personal knowledge graphs matter for AI agents. An agent that only finds one note may answer one question. An agent that can follow relationships can explain how the note fits the larger situation.
Search Is Not The Same As Understanding
Semantic search and vector search are useful. They help find material that is meaningfully close to a query. But search alone does not always show why the result matters. It might find a note about pricing without knowing that the note belongs to Maya, legal review, and Friday's promised draft.
| Question | Plain search | Connected memory |
|---|---|---|
| Who owns this? | May require reading several notes. | Person and project links are explicit. |
| Why did we decide this? | Finds candidate notes. | Shows decision, source, and related constraints. |
| What is still open? | Depends on exact wording. | Open promises and follow-ups are connected to people and projects. |
What Dossier AI Is Building Toward
Dossier AI is a private memory you talk to. When you brief what happened, the useful details can become connected records: people, promises, projects, decisions, ideas, and follow-ups.
That connected memory is what future agents will need. A good agent should not just say, "I found a note." It should be able to say what the note connects to, why it matters, and where the claim came from.
You can also see a more visual version of this idea in the connected memory demo.
Keep Reading
FAQ
What is a personal knowledge graph?
It is a connected map of people, projects, decisions, promises, preferences, and source evidence.
Why does it help AI agents?
It helps an agent retrieve related context and explain how one detail connects to another.
Is the graph public?
No. Dossier is private memory. It only holds what you choose to brief.
Dossier AI is live on Android. Tell it what happened. Ask about it later.