Private beta · early access

The memory your AI tools don't have.

Every AI session your team runs produces decisions, discoveries, patterns. Then it vanishes. Brainiac imports those conversations, turns them into team and project knowledge, and keeps them ready for the next chat.

// the problem

Your team's AI conversations are invisible.To each other, and to the next session.

01
Isolated chats

Conversations with AI assistants are your new intellectual capital, and they live in isolated windows.

02
Same answers, every teammate

Every person on the team rediscovers the same decisions, shortcuts, and constraints.

03
Onboarding is a shrug

"Here are the tools, good luck." No transmission of how the team actually works with AI.

04
grep isn't enough

Transcripts are huge and unstructured. Full-text search alone is hopeless.

// the solution

A collaborative, multi-tenant platform.Where AI conversations become team knowledge.

  • One-click import. Pull your full Claude Code, Codex, and AI session history straight from your machine.
  • Scoped knowledge items. Personal, project, or org. Fine-grained permissions on every artifact.
  • Hybrid search. Full-text plus semantic retrieval, so prior sessions actually surface when you ask.
  • Analytics dashboard. What you're asking, how much it costs, and where it compounds over time.
~/brainiac · import
brainiacimport --source claude-code
→ scanning ~/.claude/projects/
→ found 47 sessions · 2,481 messages · 1.2M tokens
→ streaming jsonl · chunking · embedding...
████████████████████░░░░ 84%
✓ imported · indexed · ready to search
brainiacsearch "why did we pick the monthly pricing tier"
// 3 results · 0.18s
@pricing·2026-03-12
Monthly won over usage-based because onboarding friction was killing the trial funnel. Finance signed off on the margin model.
@growth·2026-02-04
Churn models show annual cohorts are 2.1x stickier, but the monthly tier is the on-ramp. Don't remove it.
@ops·2026-01-20
Doc in the handbook covers the call. Revisit after Q3 once the enterprise tier data lands.
// key features

Six things Brainiac does.So you stop losing context.

Organizations, Teams, Projects

Multi-tenant hierarchy with org owner / admin / member / guest, team groupings, and projects scoped private / team / organization.

❯_
Chat sessions

Capture Claude Code, Claude CLI, Codex, Cursor, OpenAI, Ollama, and whatever you plug in next. Full-text search, filters, token tracking.

Knowledge base

Three scopes, three sources. Tagging, archiving, point-to-point sharing with users or teams. Hybrid full-text and semantic retrieval.

Imports

Import months of sessions at once. Scans local directories, streams .jsonl incrementally, resumable on failure.

Analytics

Tokens, sessions, projects, knowledge. Trends, provider mix, peak days, top lists across today, 7d, 30d, 90d, all-time.

Enterprise-grade security

Magic-link and Google OAuth. Row Level Security on every table. Policies tested against cross-org leakage.

// brainiac mcp

Your AI tools talk to Brainiac.Directly. Over MCP.

One endpoint. Every assistant you use. Claude Code, Cursor, ChatGPT, and anything else that speaks MCP can read from and write to your team's knowledge without you leaving the chat.

connect · one line
// Claude Code
$claude mcp add brainiac --url https://mcp.brainiac.app
// Cursor / ChatGPT / any MCP client
$paste the same URL
// during a session
search your team's knowledge
pull project history and decisions
ingest the finding back
scoped to your permissions
Search from inside your session

Ask Claude Code or Cursor a question, they query Brainiac and hand you the team's prior answer. No tab switch.

Pull context on demand

Past decisions, project history, and scoped knowledge items get surfaced the moment your assistant needs them.

Ingest findings back

Anything the AI concludes can be saved to Brainiac in the same breath, tagged and scoped automatically.

Scoped to you, not the world

Results only include what your account can see. Permissions are enforced end-to-end, every request.

// how it works

Four steps.Nothing that needs a Friday afternoon to set up.

  1. 01
    Create an org

    On first login, or accept an invite. Each user gets a personal org by default.

  2. 02
    Import your history

    Pull months of AI session history, from any assistant your team uses, in one click.

  3. 03
    Organize and invite

    Create projects, set visibility (private, team, or org), invite the rest of the team.

  4. 04
    Compound weekly

    Every new session deepens the base. Onboarding, research, and prior decisions become searchable.

// who it's for

Built for teams who treat AI conversations seriously.As real output, not throwaway chats.

Teams shipping with AI
Engineering, research, design, marketing, ops. Any team whose best thinking now happens alongside an assistant.
Leads who want visibility
See what your people ask AI, what works, and where your spend compounds into durable knowledge.
New hires
Absorb the team's way of working with AI from day one, not over the next six months.
Organizations protecting IP
Treat AI transcripts as the asset they are. Keep them scoped, searchable, and auditable.
// faq

Questions we get often.

  • Import from Claude Code, Claude CLI, Codex, Cursor, OpenAI, and Ollama today. The MCP connects Claude Code, Cursor, ChatGPT, and anything else that speaks MCP.
  • Search your team's knowledge, pull project history and past decisions, and ingest new findings back into Brainiac. All of it without leaving the chat.
  • No. Every request carries your identity, so results only include what you personally can access. Cross-org and cross-project leakage are blocked at the database layer.
  • Only what you upload. The importer reads from your local filesystem and you pick what goes into Brainiac.
  • Yes. Each user gets a personal org on signup with the full feature set, including the MCP.
  • Both. Full export as JSON, and a full cascade delete on account removal. GDPR-aligned, never locked in.
  • Free during private beta. After that, per-seat with a generous free tier for solo users and small teams.

Stop losing what your AI already knows.

Turn every session into team capital. Private beta is open, and we're onboarding a handful of teams this quarter.