Skip to main content
EN
Get started
Runs on your own machine

How the AI ticket to merge pipeline works

The Volath AI ticket to merge pipeline has 12 board statuses in its default workflow. You write a ticket, an AI agent does the development work in your git repository, and you stay in control from the Board. It runs on your own machine. Your code, tickets and settings live in a local SQLite database, or in your own MySQL or MariaDB.

The pipeline

From To Do to Done

In the default workflow every ticket moves through these board statuses, in this order. Not Ready and Ready are the two outcomes of the analysis. Each project can change the steps in Workflows.

  1. To Do

    IA
    Nothing runs yet.
    You
    Write a ticket from the Board. Tickets created by the Project Planner also start here.

  2. Analyzing

    IA
    An AI agent analyzes the ticket. The provider and model can be set globally, per project and per ticket.
    You
    Optionally attach documentation and give the ticket an Agent in its options.

  3. Not Ready

    IA
    The analysis ends with the ticket either Not Ready or Ready.
    You
    Read the analysis and the validation criteria in the ticket drawer.

  4. Ready

    IA
    The analysis ends with the ticket either Not Ready or Ready.
    You
    Read the analysis in the ticket drawer. Policies can hold the next steps until you approve.

  5. In Development

    IA
    The AI writes the code in your git repository and runs external review tool rounds.
    You
    Follow the activity log and the AI cost in the ticket drawer.

  6. Development Done

    IA
    The AI opens a pull request.
    You
    See the pull request in the ticket drawer, or on the Branches page.

  7. Waiting for CI

    IA
    CI failures are fixed automatically.
    You
    Watch the ticket on the Board.

  8. Code Review

    IA
    Review comments are answered automatically.
    You
    Review the pull request yourself. Volath can open pull requests with your token, so check what it opens.

  9. On Staging

    IA
    Volath deploys to staging.
    You
    Set the environments in Project settings. A policy can hold the merge into staging until you approve.

  10. On Testing

    IA
    Playwright drives a browser against the staging environment and stores screenshots.
    You
    Read the live-test results and screenshots in the ticket drawer.

  11. Awaiting Merge

    IA
    The merge action of the workflow either merges automatically or waits for a person.
    You
    When the workflow waits for a person, this is where you decide. A Gate step or a policy can also hold the ticket until you approve.

  12. Done

    IA
    Memory learns lessons from merged tickets, unless learning is turned off.
    You
    Review, edit or delete the lessons on the Memory page. The Done column is hidden on the board by default, and learning can be turned off per project and per ticket.

Human control

You decide where the pipeline holds

By default the pipeline runs on by itself. Turn on policies to keep a human decision where it matters.

  • Approval gates

    Require approval before the AI starts developing, before the pull request is opened, or before the merge into staging. A held ticket can be approved or rejected in its drawer, by the chat assistant, or through the API.

  • Needs attention

    A held ticket shows in Needs attention and in its drawer. An approval.requested webhook event is sent when a step is held.

  • Policies

    In Project settings › Policies, auto-run only tickets with a label such as safe. By default the pipeline runs on by itself, and nothing is held until a policy is turned on.

  • Budget limits

    Auto-run a ticket only while the AI spend on it stays under a limit you set. A budget.exceeded webhook event can report it to your other tools.

  • Change the steps

    The statuses are the steps of a workflow. Edit them for every project or for one, add decisions, parallel steps and gates, or start from the catalog.

  • Extend it with plugins

    Plugins add your own pages, automations and integrations. Every call is checked against the permissions you accepted.

Use cases

For solo developers and teams

  • Solo developers

    Write the ticket, let the AI do the development work in your own repository, and follow it on the Board. Describe a whole project to the Project Planner and get tickets with subtasks. Everything stays on your machine. Features: Board, AI development, Project Planner.

  • Teams

    Hold the pipeline where a human decision matters with Policies, give tickets an Agent or a Squad, and report events to Slack or Teams with Webhooks. Change the steps with Workflows and add your own automations with Plugins. Reviews lists the pull requests that request your review. Features: Policies, Workflows, Agents, Squads, Webhooks, Plugins, Reviews.

Take your next ticket from idea to merged code

It runs on your own machine. Your code, tickets and settings live in a local SQLite database, or in your own MySQL or MariaDB.

Get startedSee pricing