ThinkPort / How it works

Three steps. Then just ask.

Connect your data, set the rules, and use the AI your team already knows. No new tool to learn and no data to move.

Step 01

Connect your data

Link the places your company's data and knowledge already live. Everything stays in your private environment.

  • Sign in, don't migrate

    Email, Google Drive, Notion, laptops and phones connect with their usual sign-in. Nothing is copied to a new system.

  • Your private environment

    A private data bucket with a dedicated GPU, provided by ThinkPort or running on your own machine.

  • Always up to date

    Sources stay in sync, so answers reflect the latest emails, notes and knowledge.

Illustrative example

Email

Work inboxes

Connected

Google Drive

Shared folders

Connected

Notion

Team wiki

Connected

Laptops & phones

Notes and photos

Connected

Step 02

Set the rules

Decide which people and AI agents may receive which context. Start from ready-made rules and adjust them to your company.

  • Sensible defaults

    Sensitive details such as salaries, client names or health data start out withheld.

  • Per team and per agent

    Sales, legal or finance each see what they need; external agents receive only what you approve.

  • A receipt for every request

    See what was shared, what was withheld and which AI answered.

Illustrative example

Project timelines

Shared

Product specifications

Shared

Client names

Withheld

Salaries

Withheld

Step 03

Use any AI

Connect Claude, ChatGPT, Copilot or your own agents through MCP. Your team asks as usual; ThinkPort adds only the authorised context.

  • Connect once

    Add ThinkPort to the AI tools your team already uses. No new interface to learn.

  • Answers with real context

    Your World Model links people, customers, suppliers and rules, so AI understands your company.

  • People and agents alike

    Works when a person asks an agent and when agents talk to each other.

Illustrative example

Draft a status update for the Milan supplier.
Here's a short update covering the revised delivery date and the two open items…

Receipt

  • · 3 facts shared
  • · 2 details withheld
  • · Answered by Claude

Behind each request

What happens when someone asks.

It takes a moment and needs nothing from the person asking.

  1. 01

    Ask

    A person or agent asks their AI as usual.

  2. 02

    Understand

    ThinkPort finds the relevant people, projects and facts in your World Model.

  3. 03

    Check

    Your rules decide what this requester may receive.

  4. 04

    Share

    Only the approved context is passed to the AI.

  5. 05

    Record

    The answer comes back with a receipt of what was shared.

Getting started

From first call to everyday use.

  1. 01

    Talk to us

    We look at your data sources, your AI tools and what must stay private.

  2. 02

    Your environment

    We set up your private bucket and dedicated GPU, or install on your own machine.

  3. 03

    Connect and set rules

    Your team links its sources and reviews the default rules.

  4. 04

    Start asking

    People keep using their AI; you review receipts whenever you want.

Questions

Good questions, clear answers.

Do my people need to learn a new tool?

No. They keep using the AI they already know. ThinkPort works in the background through MCP.

Is our data copied or moved?

No. Sources are connected, not migrated. Your data and knowledge stay in your private environment.

Which AI tools can we use?

Any AI that supports MCP, such as Claude, ChatGPT or Copilot, plus your own agents.

Who decides what is shared?

You do. Rules can be set for each team, person and agent, and every request has a receipt.

What is the World Model?

A private map of your company: people, customers, suppliers, places and rules, connected so AI can understand context.

See it with your own data and knowledge.

We'll walk you through the three steps with your sources and the AI your team already uses.