Memory for agents
It's corrective, typed and informed
For memory layer, when the storage and semantic understanding is based on the approximation, it leads to the wrong answers, hallucinations and loss of trust for the users. Polign is aimed to scrub the old ways, and provide a memory layer that is corrective, typed and informed.
Why agents give wrong answers
When an agent gets something wrong, the cause is often what it was given, not the model. Most agents look up whatever text resembles the question. An old fact and its correction come back together, and the model picks one.
Customers get the current answer
A refund approved in chat and revoked a week later stays revoked on the phone, in email, and in every other agent. A change is recorded once and every agent sees it.
Decisions you can explain
Nothing is erased. When a customer, an auditor, or your own team asks why an agent acted, you can show what it knew on that date, what changed since, and where each fact came from.
Work survives an outage
If an agent crashes or is restarted in the middle of a task, the next one continues from where it stopped. Customers are not asked to start over.
Measured results
Who it's for
Teams whose agents talk to the same people again and again, where the facts change along the way and a wrong answer has a cost.
- Customer support: a promise made in chat holds on the phone, and a revoked one stops holding everywhere
- Voice agents: callers are remembered and never have to repeat themselves
- Sales and account teams: plans, contacts, and renewal dates stay current across every conversation
- Healthcare and financial services: you can show what the agent knew when it acted, and the data never leaves your account
- Long-running agents: work survives a crash or a redeploy instead of starting over
- Coding assistants: preferences and project conventions carry from one session to the next
Your data stays in your account
Recall runs in your cloud and stores memory in storage you already own. Your memory data does not pass through Polign.
- Your cloud: AWS, Google Cloud, Azure, or your own servers
- Your keys: with polign_db, data is encrypted before it leaves your servers, with keys you hold
- Full history: every change is kept with its date and the words it came from
- Separate memory per team or customer, each with its own access keys
- Open source under Apache 2.0, so you can read every line and are not tied to us
- Works with what your team uses: LangGraph, Pydantic AI, the OpenAI Agents SDK, LiveKit, Vapi, Claude Code, and Qdrant
For your engineering team
Polign Recall is the memory and state layer your agents call. polign_db is the database under it, built to keep everything in your own cloud storage and run on as few or as many servers as you need. An engineer can have it running in a few minutes.
Connect with us
Self-hosting is free. Production support is a flat annual agreement, and design partners start with 60 days free. See pricing.
Questions, design-partner conversations, commercial terms. connect-us@polign.com
GitHub
Recall's source, examples, and issues, open source under Apache 2.0. github.com/Polign/recall
Live demo
- polign.com/memory-demo Two agents on two nodes sharing one typed memory in a namespace of your own.
Ready to give your agents a memory?