Quick start: the desktop app#

This gets AI SprintFlow running on your Windows PC and working on a first Jira story in about fifteen minutes.

1. Before you install#

Optional: Docker Desktop (to build in containers instead of on your PC), ripgrep (faster code search) and ffmpeg (browser-proof videos for UI stories).

2. Install#

Install AI SprintFlow from the Microsoft Store (Download). Open it from the Start menu: the app starts a private console on this computer and opens it in your browser. The console listens on 127.0.0.1 only, so nothing on your network can reach it.

The first time, the browser shows a one-time setup page: create the admin account (a password of 12 or more characters). Only that link can create the first account.

3. Connect Jira, your code host and an AI model#

Open Connections and fill in the three cards. Each has a Test connection button that tells you exactly what is missing.

1. Jira: your site URL, the account's email and an API token. Use a bot account if you can, so comments and status changes are clearly from AI SprintFlow.
2. Source code host: the repository URL and an access token that can read code and open pull requests (for GitHub, a fine-grained token with Contents and Pull requests, read and write).
3. AI model: pick a provider and enter its key and model. For a model on your own hardware, choose Ollama, LM Studio or vLLM and enter its address. To use any other OpenAI-compatible service, tick Use my own AI provider and enter its URL, key, model and, if needed, the API version and extra headers.

Credentials are stored encrypted on your computer and are never shown again, only a masked hint.

4. Run your first story#

Open Runs → New run, enter a ticket key such as PROJ-123 and start it as a dry run the first time: AI SprintFlow does all the work but pushes nothing. Open the run to follow it live:

• Timeline: understand, plan, implement, verify, review, browser proof and publish, with the time and cost of each step.
• Changes: the code it wrote, file by file.
• Report and Proof: test results, the reviewer's verdict and, for UI stories, a video of the feature working.

When you are happy, run stories for real: AI SprintFlow pushes a branch, opens a draft pull request and comments on the Jira ticket. Nothing is merged without a person.

5. Let it work through your sprint#

Under Settings → Sprint automation, connect your Scrum board. When a sprint starts, AI SprintFlow queues the stories and implements them one at a time; the Sprint page shows the queue, and you can pause it at any time.

Useful settings to look at next:

• Budgets and Limits: cost caps per run, per day and per month.
• Data guard: replaces personal data and secrets with placeholders before anything reaches the AI model, and keeps files you mark as sensitive from being sent at all.
• Risk approvals: rules for risky code (for example database migrations, or files with code owners) that need a person's approval before the pull request opens.
• Jira status updates: move tickets to In Progress and In Review automatically.
• PR follow-up: AI SprintFlow acts on reviewers' comments and failed CI checks on its pull requests, pushing fixes to the same branch.

6. Free trial and license#

AI SprintFlow is free for 7 days with every feature and up to 5 seats. The banner at the top of each page shows the days left. After the trial a license is required to start new work; everything you made stays viewable. Buy a license on the Pricing page, then paste the key under Administer → License. See Licensing for how seats work.

7. Your data#

Everything stays on your computer: settings, encrypted credentials, run history and what AI SprintFlow has learned about your repositories. It sends data only to the services you connect: your Jira site, your code host and your AI provider. See the privacy policy.

To uninstall, use Settings → Apps → AI SprintFlow → Uninstall in Windows, or Settings → Uninstall from this computer in the console.

Something not working?#

Still stuck? Contact us.