🔗 On PyPI · v0.1.0 · Apache-2.0

Memory for your event streams.

Associative memory that links moments by co-presence — what was there together — not just by how much they resemble each other. The context that similarity search keeps missing.

pip install copresence
Shipped — pip-installable now Apache-2.0 Local-first
The idea

Similarity asks "what does this resemble?" — the wrong question, half the time

Vector search finds moments that look alike. But often the link that matters is that two things happened together: this error and that deploy, this spike and that user action, this note and whoever was in the room. Resemblance can't see it.

copresence records co-occurrence as a first-class signal, so you can pull up the moment that was there when something happened — even when it looks nothing like your query.

What it gives you

Recall by context, not just resemblance

🕸Co-presence linksAssociate events by what shared their moment — a signal similarity search structurally can't produce.
🌊Built for streamsDesigned for ongoing event data — logs, traces, timelines — not static document sets.
🔒Local-firstRuns on your hardware. No cloud service, no data leaving the machine.
🧩ComposableA small library you drop into your own pipeline — pair it with similarity, don't replace it.
Try it

Install and link your first moments

pip install copresence

# link moments by what was there together
python3 -c "import copresence"
v0.1.0, Apache-2.0. Early and honest about it — the mechanism is real and installable today. Full source and examples on GitHub.

Give your streams a memory

Free and open. Install it and start linking events by the context they shared.