Features / AI and MCP

Avocet works with your AI assistant. Use one to speed up the workflow, or none at all.

An assistant connected over MCP does the repetitive steps, on editable layers and keyframes. Avocet is complete without one, and we do not train AI models on your work.

The steps that repeat

You make the design and the motion. An assistant can take the work that comes after.

  1. Give it the list of formats. It adds each size next to the prime, and the layers reframe by their rules.

  2. Point it at a design you made. It marks the text, images and video that should change as fields.

  3. Give it a sheet. It makes an entry for each row and renders every size, one video per row.

  4. Ask it to read the open comments and make the changes you describe. Each one is marked done with a link to the edit.

You stay in control of it

The assistant works inside your project, on your terms.

Signs in as you

Add the Avocet server to your assistant. It signs in as you and sees the projects you can see.

Stays editable

What it makes is ordinary layers and keyframes. Open the project and change or delete any of it.

Uses your library

It places your components and uses your text styles and tokens, so the result matches the brand.

Shows its work

Every change appears in version history under the assistant's name, and you can restore the state before it.

Or do every step yourself

Avocet is complete without an assistant. Everything on this page is a tool you can use by hand.

See all features

Animate a scene with your team

More than two people can edit one timeline at once. Each has a named cursor, and every change shows up as it happens.

Repeat a layer into a pattern

One source layer becomes a grid, a ring or a path of copies. Edit the source and every copy follows.

Questions about AI and MCP in Avocet

Do I need an AI assistant to use Avocet?

No. Everything in Avocet works without one. An assistant is optional, and you decide how much of your work it touches.

What does an AI assistant speed up in Avocet?

Adding sizes from a list, turning a finished design into a template, filling entries from a spreadsheet, applying the review comments you point it at, and rendering every size and version in one run.

Can an AI assistant build a template in Avocet?

Yes. Point it at a finished design and it marks the text, images and videos that should change as template fields. Then it can make an entry for each row of your spreadsheet and render every size.

Does the assistant make flat images or editable files?

Editable files. What the assistant makes is ordinary layers and keyframes. You can open the project, move a key, change the easing or delete any of it.

Can I see and undo what an assistant changed?

Yes. Every change appears in version history under the assistant's name, and you can restore the state before it.

Can I use AI-made images or video in Avocet?

Yes. Avocet imports images, video and audio from any tool as ordinary files, so AI-made assets work like any other. An assistant connected over MCP works on editable layers and keyframes, not flat images.

What is MCP, and which assistants can connect?

MCP, the Model Context Protocol, is an open standard that lets AI assistants use tools in other applications. Claude, Gemini, Cursor, GitHub Copilot, Windsurf, Zed and any other assistant that supports MCP can connect to Avocet.

Which Avocet projects can an assistant see?

An assistant signs in as you and sees the projects you can see, and no others.

The beta is invite-only.

We are opening Avocet to motion designers a few at a time. Tell us who you are and we will send an invite when there is room.