Skip to content
stored
Sign inStart free
Your agents’ memory now lives in the cloud

Your agents’ second brain.Shared, and in the cloud.

Start freeSee how

Right now, your agents forget you every morning. You explain the schema to Claude Code on Monday and to Cursor on Tuesday. The context file lives on one laptop, drifts out of date, and the gotcha that cost you a Friday gets rediscovered by someone else next month.

Teach it once, and the organization knows. Your agents write what they learn into one cloud memory and read it back over MCP. A decision made in Claude Code on Monday is there in Cursor on Tuesday — and in your teammate’s agents too. You can see all of it, and fix anything that’s wrong.

Ask it anything. Get the answer, with receipts.

No more archaeology through old threads. Ask the way you’d ask a colleague who has been there the whole time — and every line of the answer points back to the memory that produced it and the agent that wrote it. Learn more

What did we decide about billing, and what should I avoid touching?
Memorybilling · decisions · whole org
  • 0.98Billing runs on Polar, not Stripeclaude-code · 14 Apr · decisionSource
  • 0.91Polar webhooks are idempotent by event idcursor · 2 May · convention
  • 0.84Legacy Stripe keys stay unusedsupport-bot · 9 May · gotcha
  • 0.77Plans are priced by usage, never per seatresearch-agent · 21 May · decision

Billing runs on Polar, decided on 14 April. The Stripe keys still in the repo are legacy — leave them alone.

  • Hybrid retrieval · graph + vector + wiki
  • 4 memories, each with a source agent
  • Same answer for every connected agent
  • Wrong? Edit it once, in the dashboard

Written by claude-code, read back by every agent in the org.

One remote endpoint and an organization key. Every agent you run, on the same brain.

Explain it once. Every agent knows.

Today that knowledge is scattered across context files, a vector store nobody can read, and the part only Sam remembers. stored makes it one memory your agents write to and read from, and your team can open.

  • Remember that we moved billing to Polar
    • Billing runs on Polardecision
    • Polar → webhooks → orders3 entities
    • Supersedes: Stripe checkoutlinked
    Extracting the session3 of 4

    Nobody explains it twice. Agents write what they learn as they work — decisions, conventions, gotchas — linked into entities, not dumped into a file.

  • What do I need to know before touching checkout?

    Hybrid retrieval

    Graph · vector · wiki

    • Billing runs on Polar0.98
    • Webhooks are idempotent0.91
    • Stripe keys are legacy0.84

    One call, reranked

    Answers, not archaeology. One call gets the whole picture back: a graph walk, vector similarity and the wiki layer, fused and reranked into a sourced answer.

  • That’s out of date — fix it once

    Written by cursor · 2 May

    Checkout goes through Stripe

    Checkout goes through Polar

    • Claude Code reads the correction
    • Cursor reads the correction
    • Support bot reads the correction

    Wrong once, wrong nowhere. Open any memory, see which agent wrote it, and correct it. Every connected agent reads the fix on its next call.

Connected with a remote MCP endpoint and an organization key.

Whatever your agents learn, it keeps its shape.

  • Decision
  • Convention
  • Gotcha
  • Entity
  • Relation
  • Preference
  • Snippet
  • Spec
  • Owner
  • Deadline
  • Endpoint
  • Schema
  • Runbook
  • Fix
  • Policy
  • Glossary
  • Account
  • Metric
  • Alias
  • Source
  • Tag
  • Quote
  • Secret ref
  • Reminder

stored doesn’t flatten a session into one long note. Each thing an agent learns becomes a typed memory with its own entities, relationships and provenance, so retrieval can walk between them instead of guessing — and so you can open any one of them and see exactly where it came from.

Why stored exists

What your company knows lives in one person's head, one person's laptop, one thread nobody can find. Agents made that worse: each one learns in private, then forgets by morning. stored keeps what they learn where the whole organization can reach it, correct it and build on it.

A brain you own, not a box you rent.

Your keys stay yours, every memory shows who wrote it, and the engine underneath is open source. If you ever want out, you can take the whole thing with you.

  • Embedding keyBYO
    OpenAI keyYours
    At restEncrypted
    ScopeThis org

    Your key, your bill. Embeddings run on your own OpenAI key, encrypted at rest and scoped to your organization. Rotate it whenever you like.

  • Organization API keys
    Stored asHash only
    ShownOnce
    RevokeAnytime

    Keys you can take back. Agents connect with per-org API keys. We store only a hash, show the secret once, and you revoke it in a click.

  • Memory historyEvery entry
    claude-code wrote it14 Apr
    Observed in sessionLinked
    You corrected itToday

    Nothing arrives anonymous. Every memory records the agent that wrote it and when. Open it, read its history, edit or delete it.

  • One workspace
    Owner · youFull access
    Admin · 2 people
    Member · 9 people

    One memory, clear roles. Owner, admin and member over a single shared graph. People are free on every paid plan, so invite everyone.

Open-source engine you can readBring your own OpenAI keyProvenance on every memoryExport your organization anytimeRead the engine

Start free. Never pay for a teammate.

The bill follows what your agents actually do — memory kept and calls made — so inviting the whole company to read and correct the brain costs nothing. People are unlimited on every paid plan. Compare plans

  • Solo · free10k
  • Builder250k
  • Scale1M
Plan ceilings, not benchmarks. Scale meters anything past its included bucket; Enterprise is sized to you.

One brain per organization, not one per laptop.

The workspace holds everything your agents have learned. Invite the people who should see it, connect the agents that should use it, and stop keeping the important parts in your head.

  • Every tool you use picks up where the last one left off
  • Free forever, and it starts paying off on day two
  • Your OpenAI key, your memory, exportable whenever

Learned on Monday. Useful on Thursday.

Coding agents

Your codebase stops being re-explained every session.

Claude Code and Cursor stop re-learning your codebase every session. The choices you made, the patterns you follow, and the traps you already hit become memory they can read on the next task.

  • Monday · claude-code remembersWe use Postgres 16 in prod, and all timestamps are stored in UTC. Never run migrations directly against prod.
  • Thursday · someone asks“how should I add the new audit table?”
  • Instantly · shared memory recallsthe UTC-timestamp convention and the rule to ship migrations through CI, not straight to prod.

Priced by what your agents do, not who you hire.

Every plan includes the remote MCP endpoint, the live memory graph, and provenance on every memory. Humans are free on every paid plan — invite the whole company.

  • Solo

    For solo builders

    $0forever

    No card needed

    Start free
    • 10,000 stored memories
    • 100,000 agent calls / month
    • Up to 3 members
    • Remote MCP endpoint + org API key
    • Live memory graph: browse, search, edit
    • Bring your own OpenAI key

  • Builder

    Most popular

    For serious builders

    $190per year

    Cancel anytime

    Start building
    • 250,000 stored memories
    • 2,000,000 agent calls / month
    • Unlimited members: humans are free
    • Premium reranker for sharper retrieval
    • Full provenance and edit history
    • Everything in Solo

  • Scale

    For scaling agent fleets

    $990per year

    + usage beyond included

    Start scaling
    • 1,000,000 memories included, then metered
    • 20,000,000 agent calls included, then metered
    • Usage: $5 / 100k memories · $2 / 1M calls
    • Unlimited members: humans are free
    • Multiple API keys + premium reranker
    • Everything in Builder

  • Enterprise

    For larger orgs

    Custom

    Talk to us

    Talk to sales
    • Dedicated or self-hosted data plane
    • SSO + custom contracts and SLAs
    • Volume usage commitments + discounts
    • Priority support and security review
    • Everything in Scale
Prices in USD. Members are unlimited on every paid plan. Questions? Talk to us.

The things people ask before connecting an agent.

Something missing? Write to us and a human answers.

  • One shared memory layer for your AI agents. Agents write what they learn and read it back over MCP or the API, so a decision made in Claude Code is available to Cursor the next morning — and to your teammate's agent too. You see the whole memory in a dashboard: search it, correct it, delete what's wrong.

Your agents are learning today.Start keeping it.

Start free