In short: your AI assistant can now work inside RenderThink. Claude, ChatGPT, Mistral Vibe or any MCP-compatible assistant connects to your account in a few minutes. Ask it for a render, an edit, a decoration, a moodboard or a client presentation: it does it in your projects, with your credits, and tells you the price before each run. The connector is included in every plan, free trial included.

Many of you already keep an AI assistant open all day: for a tricky email, a site report, a palette idea. What was missing was the link to your everyday work, your projects and your images. Now it is there. This article explains what the connector does, how to set it up, what you can ask it and how you stay in control.

What changes

AI assistants can already create images. But an image invented from a few words is not the render of your project: the room changes, the furniture changes, the materials change with every attempt. Connected to RenderThink, your assistant no longer guesses: it starts from your 3D capture, goes through the generation pipeline dedicated to interior design that you already use in the Studio, and files the result in your project.

In practice, you write "render the kitchen view of the Laurent project in 2K". The assistant finds the project and the image, tells you what it costs, waits for your go-ahead, starts the render, and the image shows up in the app right away, next to its versions. If you have RenderThink open alongside, you see it arrive live.

Which assistants?

  • Claude;
  • ChatGPT;
  • Mistral Vibe, formerly Le Chat;
  • and any assistant that accepts custom connectors, also called MCP connectors.

MCP (Model Context Protocol) is a standard plug that AI assistants are adopting one after another. You don't need to know more than that: your assistant just has to offer to "add a connector".

Set it up in a few minutes

It all starts in the app, in My account, under AI assistants. There you'll find the connector address, https://mcp.renderthink.com, a button to copy it, and the steps for each assistant. Do them from a computer, on your assistant's website or desktop app.

In Claude

  1. Open Customize, then Connectors.
  2. Click Add, then Add custom connector. Name it RenderThink and paste the address.
  3. Click Connect: RenderThink opens, and you authorize Claude.

On Claude's free plan, only one custom connector is possible.

In ChatGPT

  1. Open Plugins, click +, then choose to add a custom MCP server.
  2. Name it RenderThink, paste the address and choose OAuth for authentication. Accept the warning, then create it as a plugin.
  3. Install the plugin: RenderThink opens, and you authorize ChatGPT. Then, in a new conversation, type @ and pick RenderThink.

Adding an MCP server depends on your ChatGPT plan. In a company workspace, ask an admin.

In Mistral Vibe

  1. Open the Connectors page from the sidebar.
  2. Click + Add Connector and switch to the Custom MCP Connector tab. Name it RenderThink and paste the address as the server URL.
  3. Click Connect: RenderThink opens, and you authorize Vibe. Then, in a conversation, click + (or type /), choose Tools and turn RenderThink on.

Only an admin can add a custom connector: on a personal account (Free, Pro or Student), that's you; in a team, it's the team's admin.

With another assistant

Look for the Connectors part of its settings, add a custom connector with the address, then sign in to RenderThink when asked. And if you can't find it, ask the assistant: the AI assistants section gives you a ready-made message to copy, so it can walk you through its own interface.

The authorization

When you connect, RenderThink opens a page that clearly states which assistant is asking for access, on which account, and what it will be able to do: view your projects, change them, run paid generations with your credits. You authorize, or you decline. An email confirms every new authorization.

What you can ask it

Almost everything you do in RenderThink every day. A few examples, written the way you would talk to a colleague.

Your renders and edits

  • "Render the three views of the Martin project in 2K." It runs one render per image, up to ten at once in the same project.
  • "On the kitchen render, make the fronts matt sage green and remove the chair on the right." It traces the areas to change on the image itself; the rest stays put.
  • "The worktop is solid oak, not laminate." If it sees that an element is shown badly in your capture, it can say so before the render, like the Studio's notes.
  • "Enlarge the living room view for an A3 print." It runs the upscale.

Decoration

  • "Decorate the living room in a Scandinavian style, without overdoing it." It offers you RenderThink's styles and palettes, you decide, then it runs.
  • "Do the same decoration on the two other views of the room." Same objects, same settings, each seen from its own angle.

Your moodboards

"Put together a moodboard for the master bedroom, terracotta mood." It builds the board with you, step by step: your project images, furniture and materials from the catalog that it shows you before adding them, a color palette. You watch the board fill up in the app as you go.

Your client presentations

  • "Build a presentation for Mrs Laurent with the kitchen renders, before and after." It composes the presentation in your brand, with the images you pick together. Composing it costs no credits.
  • "Publish the link, with a password, valid until the end of November." It gives you the link to send to your client, who opens it without an account.
  • "Export it as a PDF." It gives you the file to download.

The rest of your account

  • Your projects: create one, find an image, or import images through a direct link to the file. In ChatGPT, it also imports the images you attach to the conversation.
  • Your images in full resolution: ask it for a project image, and it gives you its download link.
  • The trash: it moves there what you ask, once you agree, and can restore it.
  • Your balance and notifications: "how many credits do I have left?", "what happened on my projects this week?".
  • Support: if a render isn't right, it can write to us for you, with the right images attached, and follow up on the answer.

The price before every run

This rule has no exceptions: before any paid generation, the assistant tells you the quality and the price in one sentence, for example "in 2K, that's 55 credits, shall I go ahead?". Nothing starts without your yes. Generations use your credits at the same rates as in the app, no more, no less.

The connector itself is included in every plan. It also works during the free trial, and the images it runs carry the trial watermark, just like in the app. Your assistant's own subscription is paid to its publisher.

You stay in control

The assistant acts on your behalf, never without you. Everything is set in My account, under AI assistants:

  • you see every authorized assistant, when it was last used and what it has spent;
  • you can give it a credit limit, per day or per month; once the limit is reached, its generations wait for the next period;
  • Remove cuts off its access at once; to reconnect it, you'll have to authorize it again.

With a shared subscription, the owner decides, member by member, whether AI assistants are allowed: turning them off immediately cuts off the ones the member had already connected.

And a few safeguards that don't depend on your settings: the assistant only sees what you see in the app, and it can't pay, change your subscription or empty the trash. Anything permanently deleted is deleted in the app, by you.

Good to know

  • To work, the assistant receives your texts and thumbnails of your images: they go to it, like everything you share with it.
  • Decoration dresses a room that is already furnished: it does not furnish an empty room, with or without an assistant.
  • An image attached to the conversation only reaches RenderThink from ChatGPT. With another assistant, drag it into the project in the app, or give it a direct link to the file.
  • A generation usually takes one to three minutes. The image arrives in your project as soon as it's ready: ask the assistant how it's going, and it shows it to you.

What about the API Platform?

They are two different doors. The connector works inside your account, with your subscription: it is made for you and your assistant. The API Platform and its MCP server are for developers who build RenderThink into their own tools, with a key and a separate balance.

Your turn

The easiest way is to try it: open the AI assistants section, connect your assistant, and ask it to show you your latest projects. The rest follows naturally. All the details are also on the AI assistants page.

This is new, and the way you use it will shape what comes next: if your assistant stumbles on a request, or if you find a use we hadn't thought of, let us know.