Agent memory that learns

With Hindsight, your agents remember your users and get better at their jobs over time. Open source, MIT licensed.

View on GitHub Request a demo

Running agents at enterprise scale? Hindsight Enterprise →

The fundamentals

Everything you need from agent memory

Per-user memory: Every user gets persistent context. Preferences, history, and decisions stay separate.
Cross-session persistence: Context survives session boundaries. Pick up where you left off, weeks later.
Fast memory recall: Parallel search returns the most relevant memories in under 100ms.
Works with any LLM: Model-agnostic memory layer. Swap LLMs without losing what your agent learned.

Join the Hindsight community on Slack

Share what you're building, ask questions, and get help directly from the team.

Join Slack

A few of the teams building with Vectorize

Agents make mistakes.
Yours learn from them.

Every other memory system stores facts and retrieves them. That works until your agent encounters the same problem twice and makes the same mistake both times.

Learns from mistakes: When a tool call fails or a user corrects your agent, that becomes an experience. Next time, your agent knows what went wrong.
Detects patterns automatically: The reflection layer synthesizes individual facts into consolidated knowledge. Patterns emerge from data, not from manual tagging.
Builds judgment over time: Your agent doesn't just get more data. It gets better judgment. Curated mental models guide common situations.

The learning difference, measured

Retrieval accuracy on every benchmark with a published comparison: Hindsight against the next-best system on that dataset.

Hindsight Next best system

LongMemEval-S: 94.6%
74.0%
LoComo: 92.0%
80.3%
PersonaMem: 86.6%
84.4%
PrecisionMemBench: 85.7%
no published comparison
LifeBench: 71.5%
61.0%
BEAM · 10M tokens: 64.1%
40.6%

All benchmarks

Integrations

Memory for the agents you already use

One command gives every coding agent on your machine its own project memory. Personal agents and your own apps plug in just as directly.

terminal

$npx @vectorize-io/hindsight-coding-agents install all

# every detected coding agent, wired natively; memory builds in the background as you work

Coding agents: 20+ harnesses, one install; Claude Code, Codex CLI, Cursor CLI, opencode, Copilot CLI and more, each wired to a per-repo memory bank built from git history and past sessions. On AMB’s sdebench, Claude Code needed 0.85 → 0.36 corrections per task with Hindsight, at lower cost.; Coding agents
Personal agents: Hermes and OpenClaw; Hermes and OpenClaw install Hindsight as their memory provider: every turn recalls what matters before answering and retains what was said afterwards. The Hermes desktop app sets it up in Settings, no terminal.; Hermes and OpenClaw
Your apps: SDKs, MCP or plain HTTP; Three methods (retain, recall and reflect) from Python, TypeScript or Go, over MCP or plain HTTP. Extraction, entity linking, four-way retrieval and reranking happen behind them.; Quick start

Frameworks, MCP servers and tools are in the Integrations Hub

Run it your way

The same memory engine, API and MCP server, wherever your agents run.

Open source

Hindsight 1.0, MIT licensed, on Postgres. Run it on your own infrastructure with a single Docker command.

View on GitHub

Hindsight Cloud

Fully managed and pay-as-you-go on usage. We run the infrastructure, scaling, backups and upgrades.

Start free

Hindsight Enterprise

Hindsight deployed for your organization, in our cloud or yours, with the controls your security team will ask for.

Private cloud, BYOC on AWS, Google Cloud or Azure, or on-premises
Single-tenant isolation and private networking
SSO, Memory Defense, audit logs and SIEM
A shared Slack channel, contracted support and SLAs up to 99.95%

Plus Hindsight Memlake, our Rust memory engine on object storage, for memory that grows into the billions.

Explore Enterprise Talk to sales

Agents are evolving. Memory should too.

Most agent memory systems were built for chatbots. Those days are gone.

Agents that work together, attend meetings, learn from each other, and take on increasingly complex tasks. They need memory that compounds across every agent and every user. Retrieval-only systems won't get you there.

What compound memory looks like

01Shared context across agents: Agent A learns a user preference. Agent B applies it automatically.
02Memory that improves itself: The reflection layer consolidates raw observations into reusable knowledge.
03Per-user across every session: Months of context, instantly recalled. No retrieval lag, no cold starts.

Need help building with agent memory?

We offer hands-on implementation services, team training, and architecture consulting to get your agents learning faster. Whether you're integrating Hindsight into an existing system or starting from scratch, we'll work with you directly.

Book a call View on GitHub