> ## Documentation Index
> Fetch the complete documentation index at: https://docs.eluu.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Building a colleague

> Build an Eluu colleague in minutes — set their name, role, personality, skills, integrations, and starting memory, then put them to work.

Building a colleague takes a few minutes. Click **Add colleague** in the app and fill in five things. You can change any of them later.

## Identity

Give them a name and a role. The role is the most important field — it tells the colleague (and you) what they own.

Don't write `sales` as the role. Write something a person could read and know what to do:

> A senior sales operations colleague who runs morning pipeline reviews, drafts outreach for active deals, and flags customers we haven't talked to in two weeks.

The more specific you are, the better they perform. A short role gives you a vague colleague.

You'll also pick whether the colleague is **Personal** (just you) or **Shared** (your whole team). Default is Personal. You can flip it later, but flipping a colleague to Shared makes their full history visible to everyone.

## Personality

How they talk and how they refuse. Cover:

* Voice and tone — formal, casual, dry, warm.
* How they should structure responses.
* What they should never do — `never send email without showing me first`.
* Their default workflow when you don't specify one.

Keep it human-readable. The colleague reads it as instructions every time.

## Skills

Pre-packaged workflows. You can:

* Pull skills in from GitHub by URL.
* Pick from your team's skill library.
* Or just ask the colleague to find and add one themselves.

See [skills on a colleague](/colleagues/skills) for the full picture.

## Custom commands

Slash commands specific to this colleague. If Sofia has a `/morning-brief` command, typing it in any chat with her runs that workflow. Useful for the things you ask for repeatedly.

See [commands](/colleagues/commands).

## Integrations

Toggle which connected tools the colleague can use, and set the trust level for each one (Auto, Review, Block). See [integrations on a colleague](/colleagues/integrations).

## Memory

You don't configure memory directly when building a colleague — it builds on its own as you work together. But the colleague's [hard disk](/concepts/hard-disk) starts empty, and the more you talk to them, the more they remember. See [memory](/colleagues/memory) for how to nudge what gets remembered.

## How many colleagues to build

Most teams settle on 3-5 colleagues, each owning one role. You add more as you grow into bigger workloads. Don't pile every responsibility on one colleague — you'll get sharper output from five focused colleagues than one general one.

## Where to next

<CardGroup cols={2}>
  <Card title="Add skills" icon="blocks" href="/colleagues/skills">
    Import from GitHub or pull from your library.
  </Card>

  <Card title="Connect integrations" icon="plug" href="/colleagues/integrations">
    Pick what tools they can use.
  </Card>
</CardGroup>
