> ## 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.

# Make It Smart

> Chapter 3 of the Make Shapes guide: give your Shape a memory so it remembers you, the right model to run on, and skills it can actually use.

*Chapter 3 of the [Make Shapes](/make-shapes) guide.*

A great personality with no memory and a weak brain falls flat fast. This chapter gives your Shape three things that make it genuinely useful: a memory so it remembers you, the right model to run on, and skills it can reach for.

## A memory that makes it feel real

A Shape that remembers feels like a participant. One that forgets feels like a vending machine. Memory comes in two layers:

* **Short-term memory** is the recent messages in the chat you're in. It's what the Shape reasons over right now.
* **Long-term memory** is the lasting stuff it keeps about people and events, recalled later when it's relevant.

Long-term memory is what lets your Shape bring up last week's inside joke or remember that someone has a test coming up. It's on by default and builds automatically as you talk. Anyone in the chat can also steer it live: `/sleep` saves the conversation now, `/wack` clears the short-term slate, and `/reset` erases long-term memories.

The one setting worth tuning is the **long-term memory instructions**, where you decide *what kind* of details get saved (relationships, decisions, running jokes) instead of generic summaries. Full detail is in the [Memory Guide](/memory).

## The right brain: pick a model

Every Shape runs on an AI **model**. The same character feels different depending on which one. A creative model makes it loose and expressive; a reasoning model makes it sharp and careful.

There are **300+ models**, and most are free. You set a **primary** model and a **fallback** for when the primary is busy. Don't agonize over it: pick by vibe, test in chat, and swap if it feels off.

<Card title="Which model should I use?" icon="microchip" href="/choosing-a-model">
  Fast picks by what your Shape is for, free vs premium, and what the badges mean.
</Card>

## Skills it can actually use

By default a Shape talks. With **skills**, it can *do* things: make images with `/imagine`, search the web with `/web`, run code, read files you send it, and reach external tools like Notion and Spotify. You switch on the skills you want in a chat's settings, and the Shape decides when to use them.

Add the ones that fit the character. A coding mentor that runs code is great. A study buddy that searches the web is useful. Pick on purpose instead of turning on everything.

<CardGroup cols={2}>
  <Card title="What the AI can do" icon="bolt" href="/capabilities">
    The full menu of capabilities and how to switch each on.
  </Card>

  <Card title="Skill ideas and recipes" icon="book-open" href="/shapeskills">
    Inspiring ways to combine skills, with step-by-step setups.
  </Card>
</CardGroup>

## What's next

Your Shape is smart and capable. The last piece is the hardest and the most important: teaching it how to behave around other people.

<Card title="Chapter 4 · Make it social" icon="arrow-right" href="/designing-social-intelligence">
  When to talk, when to listen, and how to be good in a crowd.
</Card>
