Brainontic

How it works.

Five steps from request to a brain that knows you. No SDK to learn, no infrastructure to run. Just sources, a token, and the AI you already use.

The setup

From zero to grounded in five steps.

Create your account

Request early access. Approval lands in your inbox. Magic-link sign-in — no passwords, no friction.

YOU PROVIDE

Email · Use-case

Connect your sources

Drive, email (IMAP), Google Contacts, uploads, URLs. Brainontic ingests in the background and starts building your graph from the first document.

CONNECTORS

Drive · Email · Files · Web · Contacts

Create an MCP token

One token per AI client or agent. Name it. Set its scope. Optionally bind it to a specific agent so its access is logged distinctly.

TOKEN SHAPE

Name · Scope · Agent ID

Plug into Claude or Cursor

Paste the token into your MCP config. Brainontic appears as a server alongside your other tools. No SDK, no glue code.

YOU PASTE

ctk_•••• into mcp.json

Done.

Your AI now knows everything Brainontic knows. Every answer traces back to a source in your brain.

RESULT

Grounded · Sourced · Yours

Two ways to use it

For yourself. For your agents.

The same token model serves a solo user reading their email in Claude and a fleet of agents reading specs on a Cursor team.

Use case · Individual

You + Claude.

Your personal knowledge layer. One token for your Claude desktop client. Every conversation now starts with your full history loaded — projects, decisions, threads, drafts.

  you ──► Claude Desktop
            │
            └── MCP: ctk_personal
                   │
                   └── Brainontic ──► your sources

Best for: writers, researchers, founders, anyone whose work is their thinking.

Use case · Agents

Your agents + Brainontic.

One named token per agent. Each agent sees only what its scope allows. Every read is logged with its agent_id — full visibility into who asked what, when, and what came back.

  agent_a ──► tok_review_bot
  agent_b ──► tok_docs_writer
  agent_c ──► tok_support_qa
                │
                └── Brainontic ──► shared knowledge graph

Best for: dev teams, support teams, anyone running agents on shared knowledge.

Fail-closed by design

If we don’t know, we say so.

Brainontic refuses to answer when evidence is insufficient. It never makes things up. Every answer traces back to a source — a document, a thread, a page.

If the source isn’t in your brain yet, the answer is “I don’t know.” That’s the feature.

Audit, not opacity

We won’t read your data — and we make every access visible to you. Every background process, every retrieval, every token use is logged in your personal audit trail.

Brainontic

“Ready when you are.”

Request early access