How the Knowledge Graph Remembers Over Time
A second brain that just stores text is another folder of notes. What makes FactoryOS's brain different is the structure laid over the facts — what connects to what, what was true when, and how confident the system is about each piece. That structure is a temporal knowledge graph, and you can see it on the brain page.
More Than a Folder of Notes
The graph captures the relationships that flat documents lose. A note that says "Acme was our logistics vendor" gives you a sentence; the graph gives you Acme, your company, the relationship between them, and the date the relationship started or ended. Search a document and you get a paragraph. Open the graph and you see the whole picture — who, what, when, and what changed.
That difference is the difference between a filing cabinet and a working memory.
What the Visual Tells You
The brain page shows the graph as dots and lines, and the way they're drawn carries the information. Bigger dots mean stronger facts — things the system is more confident about. Thicker borders mean more durable facts, memories built to last. Color shows what kind of fact it is: a person, a company, a document, a decision, an event. Dashed outlines mark facts that used to be true and aren't anymore.
A large, solid, heavy-bordered dot is a well-established memory. A small, dashed, faint one is stale or one-off. You can read the brain's confidence off the screen before clicking anything.
Memories That Stay Findable
Forgotten facts don't get deleted — they fade. Every memory has a resilience score that controls how fast it loses strength when nothing new reinforces it, and old facts drift down the relevance list. The dot stays on the graph, just smaller and lighter, and the system still finds it when a related question comes up. "Didn't we already decide this?" gets an honest answer instead of a shrug.
Human memory works the same way. You don't lose the fact that you drove a particular car ten years ago — you just rarely think about it. The graph mirrors that: soft fade, no hard delete.
Facts With Expiration Dates
Every connection in the graph carries a start date, and when something changes, it gets an end date. Switch vendors and the old relationship doesn't get erased; it gets stamped with the date it ended, and the new vendor relationship picks up where it left off. Both are still there. Both are still searchable.
That matters because "we used to use Acme" is a real fact, often a useful one. A flat document either states it and falls out of date, or doesn't and forgets it ever happened. The graph keeps both states, in order, with the dates that separate them.
Stronger Memories From Trusted Sources
Confidence isn't uniform; it builds up. Every time a fact is reinforced — restated in a meeting, confirmed by a document, retrieved without correction — its strength climbs and its dot grows. Facts that arrive from sources you've marked reliable start out stronger and resist fading better.
So an idea mentioned once in a passing chat stays small and tentative on the graph. The same idea, restated across three formal documents and one signed contract, becomes a heavy, prominent dot. You can see what the brain trusts without reading a label.
Same Graph, Different Data
The graph isn't only useful for personal memory. The same structure fits anything where named things connect to other named things, and the human-memory layer — fading and confidence-building — can be turned off for material that needs to stay exactly as written.
- Company documents. Policies, procedures, org charts, and SOPs become memories, and the graph tracks what replaced what. Last year's expense policy isn't gone — it's an expired dot connected to the current one, and "what was the policy in March" still gets a real answer. - A codebase. Each function or component becomes a memory, and the graph remembers which other pieces of the code rely on it. "What else changes if I change this part" becomes a question the graph can answer directly, instead of searching every file by hand. - Legal documents. Defined terms, parties, clauses, and amendments become memories with named relationships. "Every clause that depends on the original definition of 'Affiliate'" returns a clean list, and a contract amended four times still shows every prior version.
In each case the original documents stay where they are. The graph is the index that knows what connects to what, and when.
This is the "super-power" of the graph, it leans heavy into how concepts are related, quite literally, through space and time.
Why Flat Docs Fall Short
A folder of files captures content but loses connections. You can find the contract, but you can't ask which other contracts cite the same definition. You can find the design doc, but you can't ask which products depend on the idea it describes. You can find last year's meeting notes, but you can't see how that decision was overturned later.
The graph is what closes that gap. It's the difference between knowing what your organization has written down and knowing what your organization actually knows.