Skip to main content

Quick start: from a description to a live app

This is the golden path: describe what your team needs, let the AI write it in your own repositories, review what comes back, and ship it with one click. Steps 1 to 4 run on the free hosted console, with nothing to install. Step 5 needs a Kubernetes cluster with Argo CD, because that is where your apps run.

Words you will see in the console

The site speaks of apps and agents. The console names the pieces they are made of: a web interface is a microfrontend, a backend or an agent is a service, and a functional domain groups the ones that serve the same business capability. An agent is a service that calls the AI model you chose. See Build with AI.

1. Create an account​

Open console.flightroom.dev and register with email or Google. You are asked to create an organization, the tenant that owns your projects, and the project wizard opens inside it.

Details: Get started with Flightroom

2. Set it up once: a Git provider and an AI​

These two steps are for whoever administers the project. After them, every member can build.

  1. Connect your Git provider. In Settings → Resources → Code Repositories, connect GitHub, GitLab or Azure DevOps. The AI writes into repositories it creates there.
  2. Add an AI. In Organization settings → AI, choose Add AI and enter your own credentials: Claude through Anthropic, Azure AI Foundry or AWS Bedrock, GPT through Azure OpenAI, or Gemini. Only organization owners and admins can do this.

Details: Connect GitHub · AI models

3. Describe it​

  1. Open Domains in the sidebar and create a New domain, for example Expenses.
  2. On the domain's Edit with AI tab, write what the team needs, the way you would explain it to a colleague. For example: "An agent that checks each expense claim against the travel policy and flags the ones over the limit before approval."
  3. Press Generate proposal.

Because the domain is empty, the AI first proposes the components the request needs, at most three microfrontends and three services. Edit the names, slugs, repositories and templates, untick what you do not want, and confirm. Flightroom creates each repository from its template, CI pipeline included, and the AI writes a first version on a branch of its own.

Details: Generate a domain with AI

4. Review it​

The proposal lands on an ai/… branch. Your base branch is untouched. From here you can:

  • keep talking to the agent until the change is right: each answer replaces the proposal on the same branch;
  • Open a pull request (a merge request on GitLab) so engineers review it where they always do;
  • Apply to the base branch directly, where branch protection allows it;
  • Discard it.

Protect the base branch and every AI change goes through review. The agent never runs code: your CI checks each branch it pushes, as it checks a colleague's.

Details: AI changes · Guardrails

5. Deploy it​

Once the change is merged, open Deployments, pick the environment and press Deploy. Flightroom commits the manifests and Helm values to Git, your Argo CD syncs the cluster, and the app answers on its own URL in that environment, with TLS.

This needs a one-time setup by a project admin under Settings → Delivery: a GitOps repository, your Argo CD, and the domain the environments are exposed on.

Details: Deploy to Kubernetes

Where to go next​

  • Core concepts: organizations, projects, domains, environments and deployments, and how they relate.
  • Console assistant: press ⌘J and ask the console to create environments, check versions or start a deploy.
  • Microfrontend quick start: the path for a web interface served by Flightroom and composed with Module Federation.
  • Self-hosting: run Flightroom on your own Kubernetes, with your single sign-on and your models.

If Flightroom is useful to you, a star on GitHub helps the project get found.