Team Brain:
shared AI memory for your whole team

Persistent AI memory at the individual level already saves time. When your AI recalls the pool exhaustion fix you documented last week, that's compounding. But it only compounds for you.

Teammate opens a new session on a different branch. Hits the same issue. Their AI starts from zero — because the lesson lives in your Brain, not theirs. The knowledge your team already paid to learn isn't shared. It's siloed in individual instances that never talk to each other.

Multiply that across five developers, six months of accumulated lessons — and you're looking at significant duplicated debugging effort that would disappear entirely if the team's AI had access to a single shared knowledge base.

A Team Brain is a shared cachly instance — one Brain, one lesson store, one Redis namespace — that every developer on the team points their AI tools to. When anyone learns something, it lands in the shared Brain immediately. When anyone starts a task, smart_recall checks the shared Brain first.

# Developer A — Claude Code — Monday
learn_from_attempts(
  topic: "auth:redis-pool-exhaustion",
  outcome: "success",
  what_worked: "set maxRetriesPerRequest: null in ioredis config",
  ...
)

# Developer B — Cursor — Tuesday
smart_recall(query: "configure redis for auth service")
// → Brain hit (confidence 0.91):
//   PATTERN: auth-service redis pool — set maxRetriesPerRequest: null
//   Source: team Brain · learned yesterday

Developer B's AI knows what Developer A's AI learned — without either of them coordinating, writing docs, or updating a wiki. The knowledge transferred automatically, the moment it was captured.

Team Brain works because cachly is a Brain layer under every AI tool, not a feature of any one tool. Developer A uses Claude Code. Developer B uses Cursor. Developer C has the VS Code extension. Developer D runs OpenClaw agents. All four point to the same instance — all four read and write to the same lesson store.

The OpenClaw Brain Bridge means automated agents contribute to the Team Brain too. A CI pipeline that fixes a flaky test can teach the Brain what the fix was. The next time a human developer hits the same flakiness, their AI already knows the answer.

In a team setting, confidence scores work differently than they do for a single developer. When three different developers hit the same issue and all apply the same fix — each confirmation raises the lesson's confidence independently. By the time it reaches 0.95, you know it's battle-tested.

When something changes — a library upgrade invalidates an old workaround — the confidence erodes as team members run into failures. The lesson self-corrects without any manual curation. Your team's shared Brain stays current because the team itself keeps it honest.

Setting up a Team Brain takes two minutes. One team member creates a cachly instance and gets an instance ID. That ID goes in the team's shared config — environment variables, a dotfile committed to the repo, or a secrets manager. Every developer and CI job that connects to cachly uses the same ID.

# .env.shared (or secrets manager)
CACHLY_INSTANCE_ID=8e03addd-a2d9-406e-bcbb-d6d8c938a3d0

# Claude Code — .claude/settings.json
{
  "mcpServers": {
    "cachly": {
      "command": "npx",
      "args": ["-y", "@cachly-dev/mcp-server"],
      "env": {
        "CACHLY_INSTANCE_ID": "8e03addd-a2d9-406e-bcbb-d6d8c938a3d0"
      }
    }
  }
}

The Brain tier determines capacity — lessons stored, recall speed, team member count. The free tier covers small teams. Pro and Business tiers add higher limits, team namespacing, and audit logs so you can see who learned what and when.

The direct savings are the easiest to measure: fewer hours re-debugging issues that were already solved. But the harder-to- quantify value is onboarding. A new developer on a team with a mature Team Brain starts with months of accumulated knowledge already in their AI. The institutional knowledge that usually lives in senior engineers' heads — or in a Confluence page nobody reads — is surfaced automatically at the right moment.

That's the thesis: compounding knowledge at the team level. The Brain gets smarter as the team works. Every day you use it, the next day costs less.