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

# Quickstart

> Start with meetings, then let your own agents write into Atlas, then connect a live gateway when you need live tool calls.

## The shortest path to value

Most teams do not need the full gateway on day one.

The practical rollout looks like this:

### 1. Start with meetings

Use Atlas as the meeting agent first.

That gets your team comfortable with:

* a live agent in the room
* generic meeting help
* shared canvas output
* built-in meeting recap and history

### 2. Bring your own internal agents

Once meetings feel natural, let your own internal agent write into Atlas through the [Workspace API](/atlas/workspace-api).

That is a clean fit for:

* Claude Code
* Codex
* Cursor-based internal agents
* OpenClaw-style local agents
* your own internal automation

Those agents can use their own prompts, skills, and repo context to:

* keep shared skills in sync
* update shared agents
* create debriefs and publish them into Atlas with the workspace key

### 3. Connect the gateway

Connect a live gateway when you want Atlas to call your tools during the meeting.

That is the right step when Atlas should:

* query private systems live
* run approvals or async jobs
* call workflows while the meeting continues

## Which path should I use?

Use the [Workspace API](/atlas/workspace-api) when:

* the work already happened elsewhere
* your own agent already knows what to write
* Atlas just needs the clean result

Use the [Gateway](/gateway/overview) when:

* Atlas should call tools live in the room
* Atlas needs action schemas, approvals, or job state
* the meeting itself is the place where the tool call should happen

## First success in one session

If you want a clean first run:

1. sign in and open the Atlas dashboard
2. start one meeting with a built-in agent
3. get comfortable with the recap, transcript, and canvas history
4. generate a workspace API key
5. ask Claude Code, Codex, or your own internal agent to upsert one debrief through the [Workspace API](/atlas/workspace-api)
6. only then wire a live gateway

If you already know you want a live gateway first, go to [Dashboard Setup](/atlas/dashboard-setup).

## What coding agents should read

If you are pointing a coding agent at Atlas, the clean reading order is:

1. this page
2. [Workspace API](/atlas/workspace-api)
3. [Build With An Agent](/gateway/build-with-an-agent)
4. [Debrief Templates](/gateway/debrief-templates)

That gives the agent the product model first, then the write path, then the live gateway contract.

## What is coming next

The near-term direction is straightforward:

* better built-in post-call work, starting with cleaner recap and artifact flows
* smoother setup paths for teams that want to get value before they wire a full gateway
* more guided gateway setup for common internal agent runtimes and enterprise deployment patterns where demand is clear

Over time, we also expect to add more opinionated, one-click gateway starters for common setups, including OpenClaw-style local agents, personal agent deployments, and enterprise-specific environments where that support meaningfully reduces setup time.