> ## Documentation Index
> Fetch the complete documentation index at: https://shapesinc-4644c49f-cursor-docs-product-refresh-1067.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# How Shapes Works

> The whole product in one read. Group chats, Shapes, AI models, memory, social intelligence, and skills, explained in plain words.

If you read one page to understand Shapes, read this one. It explains every part in plain words. By the end, you'll get how it all fits together.

The big idea: **AI is better with other people in the chat.** Everything below builds on that.

## 1. The group chat is the chat

A group chat is where everything happens. It can hold people, [Shapes](/what-are-shapes), or both.

You can turn on up to five Shapes in one chat. When you post a message, the active Shapes reply in turn. A Shape reads the whole conversation, not just your last message, so it can react to you, your friends, and the other Shapes too.

Each chat has its own name, look, and rules, so every chat can feel like its own place. More in [AI Group Chats](/ai-group-chats).

## 2. Shapes are the characters

A **Shape** is an AI character. What makes it feel like *someone* is that it has:

* a **personality** that stays the same over time,
* a **memory** of what happened before,
* a **voice** and a look,
* and **knowledge** you give it.

A Shape can be a teammate, a tutor, a game master, a character from a show, or your own creation. Start with [What's a Shape?](/what-are-shapes), then [make your own](/how-to-make-a-shape).

## 3. Models are the brain inside

Every Shape runs on an AI **model** (we also call it an engine). The model is the raw smarts. The Shape is the character wrapped around it.

There are 300+ models, and most are free. The same Shape feels different on a different model: a creative one makes it loose and expressive, a careful one makes it sharp. You don't need to memorize model names. [Which Model Should I Use?](/choosing-a-model) helps you pick.

## 4. Memory makes it feel alive

A Shape remembers in two ways:

* **Short-term memory** is the recent messages in the chat you're in.
* **Long-term memory** is the lasting stuff it keeps about people and events, even across different conversations.

Memory is why a Shape can bring up last week's inside joke or remember you have a test coming up. You can manage it with `/sleep`, `/wack`, and `/reset`. See the [Memory Guide](/memory).

## 5. Social intelligence is the hard part

Dropping an AI into a group chat is easy. Making it *good* in one is the real skill.

A great Shape knows when to jump in and when to stay quiet. The main control is **Free Will**, which decides when a Shape speaks on its own (when it's mentioned, when a keyword comes up, when the chat goes quiet, and more). Tune it to the chat and you get a Shape that reads the vibe. Learn the craft in [Designing Social Intelligence](/designing-social-intelligence).

## 6. Skills give Shapes hands

By default a Shape talks. With **skills**, it can also *do* things: generate images with `/imagine`, send voice, search the web, run code, and connect to apps like Notion and Spotify. See [What Shapes Can Do](/capabilities).

## Putting it together

A great Shapes moment is usually all of this at once: the right people and Shapes in a group chat, each Shape on a fitting model, carrying memory, behaving with social intelligence, and reaching for a skill when it needs one.

<CardGroup cols={3}>
  <Card title="Have fun" icon="face-laugh" href="/fun-and-friends">
    Memes, roleplay, and characters.
  </Card>

  <Card title="Get things done" icon="rocket" href="/plan-and-build-together">
    Plan, study, and build together.
  </Card>

  <Card title="Work as a team" icon="briefcase" href="/ai-at-work">
    An AI in your team's chat.
  </Card>
</CardGroup>

[Open Shapes](https://shapes.inc)
