MemQL exists so AI can act on data you own, and keep a record of what it did.
MemQL is an open-source AI platform built on a time-series memory graph: typed data, queries and mutations, model routing, durable agent work and event-driven automations, declared in one .memql language and run on a cluster you own.
Why it exists
AI has a structural flaw: when a conversation ends, it forgets. So every team rebuilds the same fragmented stack to paper over it — a Postgres next to a vector store next to an event bus next to an LLM wrapper next to a file of retry logic — and keeps it all consistent by hand.
AI without memory doesn’t amplify you — it hands you back work.
MemQL is the part of the system that remembers. Typed data and its history, the agent work done on it, and the automations that react to it live in one time-series memory graph, on PostgreSQL, TimescaleDB and pgvector. You declare the data and the behavior in .memql files; the engine executes them on a cluster, and every write adds a version instead of overwriting one.
What it is
The engine runs the system. MemQL OS, the browser workspace served at os.<your domain>, is an application connected to it; so are the VS Code and Cursor extension and your own sites and applications. OS apps are views and controls for particular jobs, not separate engines.
A cluster runs seven node types, each a product-agnostic engine image: agent, bff, edge, identity, mcp, planner and workbench. Model calls route through policies and rules to the sources the engine declares: Anthropic (hosted chat and reasoning models), OpenAI (hosted chat, embedding and speech models), Fleet (local models on machines you own) and Subscription app (a coding app signed in on a machine you own).
A capability being implemented does not mean a new cluster has it configured: model calls need a compatible inference source, mail needs a sender, and an integration needs its own configuration. The full picture is in What is MemQL?; this page is about why it is built.
Open source
MemQL and MemQL Cockpit are open source. The repositories are public, and what they hold is what the docs on this site describe: znasllc-io/MemQL and znasllc-io/memql-cockpit.
Who’s behind it
MemQL is built and maintained by ZNAS LLC. It’s deliberately code-forward — the work is in the open, and the repository is the resume.
Status
MemQL is alpha and pre-1.0. APIs and language syntax can change between releases. The local runtime is a Docker-backed k3d cluster managed through ArgoCD, and a cloud installation follows the same GitOps topology: it is a multi-service system, not a single editor process. What is implemented and what is still design work is on status and direction, and what the platform measures about itself is on the proving scorecard — replayed from recorded responses in CI; live tier not yet run.
Get involved
Star it, browse the source, open an issue or a discussion on GitHub. For licensing questions, or just to tell us what you’re building, email legal@znas.io. No “book a demo.”