a fact that changed
Where do I work now?
You work at Company B (as of 2023-06-23). Earlier, you were at Company A on 2023-02-04.
your ai remembers
what came before.
Ask about something you said weeks ago, revisit how a decision actually changed, or see what Backstory remembers. Everything is kept as versioned graph facts on HydraDB, not a buried transcript.
Tell it once. Come back later. The graph still has the history.
real conversations
this is what it actually answers.
The same questions you can run in the app. Click a card below to try it live.
based on 3 memories
scattered across sessions
What do I know about Priya?
Priya is the tech lead on the Atlas migration (2023-01-12). She is based in the Berlin office (2023-03-08). She now owns the billing API rewrite (2023-07-02).
refusing to guess
How many services run on the 64 GB instance?
I don't have enough information from your previous conversations to answer that.
why, not just what
Why did we choose Postgres?
We switched to Postgres after an initial plan for MongoDB, and the strong consistency requirement for billing records is what settled it.
across a gap
What is my birthdate?
Your birthdate is March 12, 1995.
simple recall
What degree did I graduate with?
You graduated with a degree in Business Administration.
how it works
every message becomes a graph fact, not just tokens in a transcript.
The same path on every turn. HydraDB is the graph of record.
01
extractPull subject / predicate / object atoms from what was said.02
resolveMatch entities by exact key or alias. Sam ≠ Samuel unless told so.03
versionNew facts SUPERSEDE old ones, or CONTRADICT them. History is kept, never overwritten.04
retrieveGraph traversal from seed entities, aware of current state.05
abstainNo matching evidence, or a false premise in the question? It says so.what it remembers
four ways long term memory actually gets tested.
Each of these is a real scenario you can run against the live engine, not a mockup.
Changed plans
A person's work history changes across several conversations. Backstory tracks the update and keeps the earlier answer on record.
run this scenario →
Remembered across conversations
Facts about one colleague arrive scattered across separate sessions and get synthesized into a single answer.
run this scenario →
Knows when it does not know
Ask something that never appeared in memory. Backstory abstains instead of guessing from unrelated evidence.
run this scenario →
Decision history
Trace how a database decision evolved from MongoDB to Postgres, and why.
run this scenario →
questions
the things worth asking first.
What happens when a fact changes?
The old value is not overwritten. The new fact supersedes it, and both stay on the graph with their dates, so the current answer and the history behind it are both available. That is why the work question above can answer Company B and still show Company A.
What does it do when it does not know?
It says so. Retrieval can return related facts and the engine will still refuse if the specific thing asked about is missing, rather than answering from something adjacent. Refusing correctly is treated as a feature, not a failure.
Is my memory private to me?
Yes. Every account gets its own isolated graph, keyed to the signed in Google account and enforced on the server rather than in the browser. Nobody sees anyone else's facts, including the demo scenarios.
Why a graph database rather than a vector store?
Because the hard part is not finding similar text, it is knowing which version of a fact is current, what it replaced, and what contradicts it. Those are edges. On a vector store, superseded and current text look equally similar to a query.
Does it need an LLM to work?
No. An LLM improves how facts are extracted and how answers read, but the memory itself is graph operations. With no model configured the engine still stores, versions, retrieves and abstains correctly, and falls back automatically if a provider fails.