Launching today
Polygres
Turn your entire database into a context window for AI
106 followers
Turn your entire database into a context window for AI
106 followers
Polygres turns Postgres into working memory for AI agents. Retrieve structured rows, connected relationships, semantic matches, and full-text results through one hybrid API. All without a separate vector store, graph database, or sync pipeline. Build grounded agents over the data you already have, with relational, graph, and vector context joined at the source. Use the managed cloud or self-host with our open source components. Polygres is currently free to use, just sign up!
@evokoa always struggling with context window and optimize cost as my db grows large, and i started moving into using polygres. i was very suprised to see how my agent can queries everything a lot more efficiently in term of memory and cost after that like literally. still working on trying out the product more as the knowledge based become larger, but this looks exciting!!!
Congrats the team on the launch
@tony_nguyen25 That's awesome to hear! And our repo's are fully OSS if you ever wanna self host :D
How can I connect it to my polygres to my agent? Can I use mcp?
@1nk Hey inky! so we have a cli and skill that you can connect directly to the agent :D https://docs.evokoa.com/polygres/agent-skills
Love the launch polygres team! How exactly do you guarantee an infinite context window?
@lucapiekarski The full database remains available as external memory. For each request, Polygres retrieves and ranks the most relevant rows, relationships, and search results, then fits that information into the model’s actual token limit.
As the conversation continues, the system can fetch additional context on demand instead of trying to load the entire database into one prompt. So the database is effectively unbounded context.
We've dubbed this the infinite context window!
The one-store argument is the right one. Sync pipelines between a vector store and the source of truth are where most of these systems quietly go wrong, so removing the sync is worth more than any benchmark number you could put next to it.
The thing I would want to know: retrieving from the live database buys you freshness and costs you reproducibility.
If the agent's context is whatever the DB held at that moment, then the same question next Tuesday gets a different answer, and I cannot reconstruct what the agent was actually looking at when it made a call I now have to defend.
Anywhere there is an approval step, that matters more than latency does.
So does a retrieval come back with a receipt? Row ids, ranking, timestamp, stored next to the decision. Not for the agent. For the human who has to explain it three months later.
Arkor
How fast does the search get?
I’m going to try this for my mental state agent. It always need more context
@dylan_bryan2 That's a great idea! Happy to help you build this out you can dm me on my X.
@dustarion perfect
Calling it infinite context is a stretch, but fetching more on demand instead of stuffing everything into one prompt is the right shape. Closer to how people actually recall things anyway.