Skip to main content

​
Introducing mini

mini is an AI agent that operates as a mobile engineer on your team. You assign it a ticket and optionally a Figma file. It writes native code in your existing codebase, deploys to a real iOS or Android device, iterates until the implementation works, and opens a pull request. It is not a code completion tool or a chat assistant. mini works autonomously end-to-end — reading the ticket, writing code, running on an actual device, fixing issues, and submitting for review — without manual intervention at each step. mini is in closed beta. Request access →

​
Capabilities

​
Build (closed beta)

Assign a ticket. Attach your Figma. Get a tested pull request.
  1. Ticket — Connect Linear, Jira, Notion, or paste a task description
  2. Figma (optional) — Attach your design file as the visual spec
  3. Build — mini writes native code directly in your existing codebase
  4. Test — Deploys to a real device and iterates until the implementation is correct
  5. PR — Opens a clean, atomic pull request scoped to the ticket
How Build works →

​
Test (closed beta)

mini builds your test suite autonomously and runs it on real devices on every change — whether you wrote the code or not.
  1. Describe — Write test cases in plain English
  2. Generate — mini builds and maintains the test suite autonomously
  3. Run — Tests execute on real iOS and Android devices on every change
  4. Report — Results with screenshots, logs, and video

​
Swarm (coming soon)

Exploratory testing across your app on every commit.
  1. Push — Triggers on every commit or pull request
  2. Deploy — Fleet of agents picks up real devices
  3. Explore — Agents tap, scroll, and stress-test your app
  4. Catch — Regressions flagged before they reach users

​
Maintain (coming soon)

Dependency management with verified device testing.
  1. Scan — Monitor dependencies and detect outdated packages
  2. Update — Bump versions and resolve breaking changes
  3. Test — Verify everything still works on real devices
  4. PR — Review a clean pull request with the changes

​
Who it’s for

App agencies and studios. Managing multiple client apps simultaneously, with engineers stretched across codebases. mini handles a portion of implementation delivery — your team focuses on architecture, client relationships, and work that requires judgment. Mobile-first product teams. A production app, a backlog that compounds faster than it shrinks, and QA that can’t match shipping velocity. mini takes implementation tickets off the backlog and delivers tested PRs. Enterprises running platform migrations. React Native to native. Web to mobile. One codebase to two. mini can work across both repositories simultaneously, testing on real devices for each platform.

​
Not the right fit

mini is not the right tool for web applications or greenfield projects without an existing codebase. For architectural decisions, mini’s planning session can help you think through the best approach — but the final call on system design remains with your team. It is built for teams with production mobile apps who need to move faster on implementation.

​
How it works

mini reads the ticket, examines the codebase to understand structure and patterns, and writes code in your actual repository. It then deploys the app to a real iOS or Android device in the cloud, navigates to the relevant screen, and verifies the implementation behaves correctly. If something is wrong, it goes back, fixes it, and re-tests. When the implementation passes, it opens a pull request. Your team reviews the PR through your existing process. mini does not push to production, does not make architecture decisions, and does not bypass code review. Frameworks: Swift, Kotlin, Flutter, React Native. On-prem available: Keep your code within your infrastructure. Contact us for private deployment options.

​
Get access

mini is in closed beta. We’re working with a small number of teams and expanding access gradually. Request access →