# Alchemist: Getting Started

Create your first project on the Alchemist autonomous software-development platform and ship your first ticket. AI agents do the engineering; the production output can be a conventional app or service with no AI running inside it.

---

Alchemist generates and operates production software for you. You describe
what you want, a team of AI agents does the engineering work (planning,
implementation, testing, review, semantic merge, deployment, and
verification), and the platform hosts the result at a live URL. There is no
IDE to install and no CI to configure -- the agents run in a sandbox, commit
to your project's real GitHub repo, and the platform rebuilds and rolls out
the running app on every successful merge.

The AI does the building. It does not have to be part of what you run in
production. A ticket can just as easily produce a backend API, a
service-to-service integration, a scheduled job, an event processor, or a
data pipeline as it can a web app, and none of those need to include a
chatbot, an LLM call, or any other AI component in their runtime. Publishing
a Chipp chat agent is one thing Alchemist can build; it is not a requirement.
See [Alchemist Architecture: The Two Planes](/docs/guides/alchemist-architecture)
for how the development side (agents, tickets, review) and the production
side (whatever you actually deploy) stay separate.

## The lifecycle

1. **Create a project.** Each project is backed by a GitHub repo cloned
   from the platform template. The repo is private and owned by the
   platform org; you can eject it to your own org at any time.
2. **Open a ticket.** Tickets are how you ask for work -- "add a login
   page", "fix the bug where checkout 500s", "switch the homepage hero
   to a video". The platform routes each ticket to an agent.
3. **The agent runs.** Inside an isolated build sandbox, the agent
   plans, implements, reviews, security-checks, and documents its diff.
   You can follow along in the Jobs view, or close the tab and come
   back later -- every step is durable.
4. **The platform deploys.** When the agent's branch merges, a fresh
   build kicks off automatically. The new image rolls out behind the
   same Live URL, with a few seconds of crossover.

## Where things live

- **Jobs** -- every agent run, in progress and historical.
- **Projects** -- your repos, live URLs, and per-project settings.
- **Domains** -- custom domains attached to a project.
- **Billing** -- credit balance, plan, and invoices.
- **Settings** -- account-level preferences (email notifications, API keys).

## Next steps

- Read **[Alchemist Architecture: The Two Planes](/docs/guides/alchemist-architecture)** to see
  how the autonomous development plane relates to whatever you actually deploy.
- See **[Alchemist Production Workloads: Non-Chat Examples](/docs/guides/alchemist-production-workloads)**
  for concrete tickets (APIs, workers, scheduled jobs, data pipelines) with no chat surface at all.
- Read **[Alchemist Tickets](/docs/guides/alchemist-tickets)** to understand how each
  ticket runs as a dynamic workflow graph, how to read node statuses live, and how
  to cancel or retry. Then read **[Working with tickets](/docs/guides/alchemist-working-with-tickets)**
  for tips on writing effective tickets and replying when an agent asks you a question.
- Read **[Environment variables](/docs/guides/alchemist-environment-variables)** to see
  what the platform pre-injects (storage, email, JWT secret) and how to
  add your own.
- Read **[Deploys and domains](/docs/guides/alchemist-deploys-and-domains)** to learn
  the difference between the Live URL and a custom domain, and how
  to force a redeploy.
- Read **[API keys](/docs/guides/alchemist-api-keys)** and the
  **[MCP server](/docs/guides/alchemist-mcp-server)** to drive the platform programmatically
  or from an AI client.