# FlyMyAI Deploy

> Deploy and manage a live realtime AI avatars endpoint on FlyMyAI.

Provisions a live realtime AI avatars endpoint on FlyMyAI infrastructure and
returns a public `flymy.ai` URL you open in a browser for a WebRTC / realtime
session. Use it only when you want to deploy, launch, or host a live endpoint -
not for batch avatar workflows (transcribe, generate speech, lip-sync).

## What it can do

| Method | What it does |
| --- | --- |
| `deploy_pipeline` | Deploy a pipeline template from the FlyMyAI catalog and return a public `flymy.ai` endpoint URL to open in the browser. Currently the only supported template is `ai-realtime-avatars`. Takes an optional `name` for the deployment; provisioning takes ~30s. |
| `stop_pipeline` | Stop the active deployment for this chat. After stopping, the public endpoint URL stops serving the service. |
| `list_deployments` | List the FlyMyAI deployments owned by this chat, with their status and public URLs. |

> **Note: Catalog**
>
> The catalog currently exposes a single template: `ai-realtime-avatars` (a
> realtime AI avatars pipeline). Other pipeline kinds are not in the catalog.
> You do not write pipeline code, model loaders, or training scripts - you deploy
> the template as-is.

## How to get credentials

None - FlyMyAI Deploy is a built-in tool. It runs on FlyMyAI infrastructure
using your account, so there is no key to paste. Just enable it.

## Fields to fill in FlyMyAI

None.

## Troubleshooting

- **Template not found** - only `ai-realtime-avatars` is available in the
  catalog today. Other template slugs are rejected.
- **No active deployment to stop** - `stop_pipeline` only works after a
  successful `deploy_pipeline` in the same chat. Run `list_deployments` to
  check what is live.
- **Public URL not loading** - the endpoint is realtime (WebRTC). Open it in a
  browser; the link stops serving once the deployment is stopped.