Tiles

In Tila, everything starts with tiles. These are visual blocks on the canvas that perform specific tasks using AI.

Each tile is a mini-agent that you assemble yourself. Inside a tile, you can:

describe a task to set the direction of work,
select the required LLM model (for example, GPT-4o, Claude Sonnet 3.7, or Gemini 2 Lite),
set parameters that affect the result (image size, video style, etc.).

All fields are filled out directly in the interface — no code, quickly and transparently.

Tiles can be connected to each other, passing context and building logical chains.

What types of tiles are there?

Tila has two types of tiles:

Universal “Request” tile — simple and minimalist, with no extra fields. Suitable for quick tasks.
Specialized tiles — with detailed settings for specific scenarios: image generation, video, documents, speech, visualization, etc.

You can read more about each tile in separate articles.

Tiles can be connected to each other. This allows you to automatically use the result of previous steps in the next one.

To create a new linked tile: hover over ⊕ at the bottom of the current tile and choose the desired type of new tile.
To connect existing ones: click and hold ⊕ or • on one of the tiles, then drag the connection from one tile to another.

💡 Tip

You can use multiple connections. For example, link three tiles at once to the next one to use their context together.

All tiles (except the “New Request” tile) have a context menu with the following options:

At the bottom of the menu, you can see the number of credits used. Hover over it to see a detailed breakdown of credit spending.