The Wiki · a living knowledge graph

A wiki your agents write and read.

Every fact your agents learn becomes part of a navigable knowledge graph. Each entity gets a page. Each claim gets a source and a confidence score. And you can always ask: what did we believe about this at time T?

entity pages provenance trail confidence-scored version history
acme-corp
ENTITY PAGE
decision_makersarah-chen 0.97
headquartersaustin-tx 0.95
primary_pain → manual reporting overhead 0.91
evaluatedcompetitor-x 0.86
renewal_quarter → Q3 FY26 0.82
5 claims · 3 outbound links 12 backlinks ↗
every linked entity is its own page · walk the graph in any direction
01

Facts become pages. Links become a graph.

Agents do not write essays into a folder of markdown. They write subject–predicate–object triples: who, what relation, to what. Each subject and each object is a node. Each predicate is an edge. The result is a wiki that assembles itself, one claim at a time.

ONE TRIPLE, WRITTEN BY AN AGENT
acme-corp decision_maker sarah-chen
subject · predicate · object — the atom of the graph
THREE PAGES, AUTOMATICALLY LINKED
/wiki/acme-corp
↳ decision_maker
/wiki/sarah-chen
the edge is navigable from both ends
NO SCHEMA TO DESIGN UP FRONT
New subjects mint new pages on first mention. The wiki grows as the agent learns.
Set-valued predicates like references and tag coexist; functional ones keep a single current value.
Backlinks are free. Every page knows who points at it, so context walks both ways.
remember · MCP call
WRITE
{
  "tool": "remember",
  "args": {
    "subject":    "acme-corp",
    "predicate":  "decision_maker",
    "object":     "sarah-chen",
    "source":     "call-2026-06-18",
    "confidence": 0.97
  }
}
 1 edge written · 2 pages touched · backlink indexed
02

Every claim cites its source.

A wiki is only as good as its citations. In abagraph, provenance is not an afterthought you bolt on — every fact carries where it came from and how sure we are at write time. Click any claim and the trail is right there. Below your threshold, the agent abstains rather than guess.

CLAIM · acme-corp / decision_maker → sarah-chen explain_fact
confidence 0.97
SOURCEcall-2026-06-18 · sales transcript RECORDED AT2026-06-18T14:22Z VALID FROM2026-06-18 · still active STATUSACTIVE
▲ A competing low-confidence write (0.61, LinkedIn scrape) landed as a Candidate — it never superseded this claim and is excluded from context until corroborated.
PROVENANCE
A trail, not a vibe

Source, recorded-at, and valid-from travel with every fact. The agent can verify any claim instead of taking the model's word for it — and so can you, on the entity page.

CONFIDENCE
Knows when to say no

Writes below your threshold land as Candidates, not truth. Re-asserting at full confidence promotes them. The wiki refuses to launder a guess into a fact.

03

Every page has a version history.

Wikis keep their edits. So does abagraph — but at the level of meaning. Because storage is bitemporal, nothing is ever overwritten. When a claim changes, the old one is closed as Superseded and the new one opens. Ask as_of(T) and the page reconstructs itself as it stood that day.

HISTORY · acme-corp / contract_tier 3 versions
enterprise NOW · ACTIVE
valid from 2026-06-18 · conf 0.96 · src renewal-call
team SUPERSEDED
2026-01-04 → 2026-06-18 · conf 0.92
trial SUPERSEDED
2025-11-20 → 2026-01-04 · conf 0.88
recall_as_of · MCP call
TIME TRAVEL
{
  "tool": "recall_as_of",
  "args": {
    "subject": "acme-corp",
    "as_of":   "2026-03-01"
  }
}
 contract_tier = team · the truth on that day
WHY TWO TIMELINES

Valid time is when a fact was true in the world. Transaction time is when you learned it. Tracking both lets you audit not just what changed, but when the agent found out — the difference between a wrong-then-corrected belief and a fact that was always wrong.

No migration, no snapshot job. Time travel is how the store works, not a feature on top of it.

04

Two wikis, one engine.

The Wiki is the knowledge your agents build — entity pages, claims, and history that grow from what they learn. The product docs are the reference you read — the API, the data model, and the recipes for writing and recalling those facts. Same engine underneath, different reader.

THE WIKI · /wiki
What your agents know
  • An entity page per subject and object
  • Claims with source, confidence, and status
  • Version history and as_of time travel
PRODUCT DOCS · /docs
How to build it
  • REST + MCP API reference
  • The bitemporal data model and conflict rules
  • Recall, walk, and build_context recipes
Open the docs →
Early access open now

Give your agents a memory they can browse, cite, and trust.

Entity pages, provenance on every claim, and time travel per fact — a wiki that writes itself as your agents work. One engine, not a patchwork.

Join early access Read the docs →