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

# Quickstart

> Start using Starfire AI in minutes.

# Quickstart

This guide gets you from a new Starfire AI account to a useful first workflow without requiring you to understand every part of the platform.

## 1. Sign in and confirm your account

Open [Starfire AI](https://ai.ethanbragdon.icu) and sign in with your account.

Depending on your account state, you may be asked to complete email verification or another account-security step before all features are available.

<Note>
  If sign-in, verification, or recovery is not working, use the [authentication troubleshooting guide](/troubleshooting/authentication) rather than creating duplicate accounts.
</Note>

## 2. Start with a focused conversation

Open a new chat and describe the outcome you want. A good first prompt includes:

* the task
* relevant context
* constraints
* the format you want back

```text theme={null}
I have a NestJS API and a Next.js frontend. Help me diagnose a login request that returns 401.

First give me the most likely causes, then a step-by-step debugging plan. Do not assume I can change the authentication architecture.
```

Starfire keeps the conversation context available to the active model within the limits of that model and the current platform configuration.

## 3. Choose the right workspace

Not every job belongs in a single chat.

| You want to…                              | Start with…            |
| ----------------------------------------- | ---------------------- |
| Ask a question or work interactively      | **Chat**               |
| Keep a long-running body of work together | **Project**            |
| Search uploaded/reference material        | **Knowledge**          |
| Investigate a topic with external sources | **Research**           |
| Build or repair a software project        | **FORGE**              |
| Reuse the same AI role repeatedly         | **Agent**              |
| Build against Starfire programmatically   | **Developer Platform** |

## 4. Add context intentionally

Starfire can work with different kinds of context, including conversation history and—when enabled for your account—files, project material, knowledge sources, repositories, search results, and tool output.

More context is not always better. Add the material that is relevant to the task and keep unrelated work in separate chats or projects.

## 5. Understand models and routing

Starfire is provider-independent at the product layer. Depending on your plan and current rollout, you may be able to select a model directly or let Starfire route the request using platform configuration.

Different models can vary in:

* reasoning capability
* speed
* context window
* tool support
* vision/file support
* streaming behavior
* cost or credit consumption

See [Models & Routing](/chat/models-routing) before assuming two models should behave identically.

## 6. Use projects for serious work

For work that spans multiple conversations or files, create a project instead of treating one chat as a permanent workspace.

A Starfire project is designed to become the durable context boundary for things like:

* a software product
* a research initiative
* a documentation set
* a business workflow
* a long-running creative project

<Card title="Project OS" icon="folder-tree" href="/projects/project-os">
  Learn how projects organize conversations, files, knowledge, artifacts, and ongoing context.
</Card>

## 7. Move to FORGE when the output is software

If your goal is not just to discuss code but to **produce a project or downloadable build artifact**, use FORGE when it is available for your account.

FORGE is designed around a staged software workflow: plan, create files, validate, repair, package, and produce artifacts.

<Card title="FORGE Build Mode" icon="hammer" href="/forge/build-mode">
  See how Starfire's build workflow differs from ordinary coding chat.
</Card>

## 8. Create an agent for repeatable behavior

If you repeatedly start chats with the same role or instruction set, turn that setup into a custom agent.

Good candidates include:

* a code reviewer
* a release manager
* a documentation writer
* a research assistant
* a support triage agent

## 9. Check usage and billing context

Features and compute consumption can be tied to your personal account or an organization. Before making billing or permission changes, confirm which context you are operating in.

<CardGroup cols={2}>
  <Card title="Credits & metering" icon="gauge" href="/billing/credits-metering">
    Understand how Starfire measures platform usage.
  </Card>

  <Card title="Organizations" icon="building" href="/organizations/overview">
    Learn how team context changes access, billing, and ownership.
  </Card>
</CardGroup>

## Next steps

Once the basics make sense, the most useful next pages are [Chat & Models](/chat/overview), [Projects](/projects/overview), [Knowledge](/knowledge/overview), and [FORGE](/forge/overview).