Scooter’s Coffee runs two revenue-critical native apps. We built a custom multi-agent AI pipeline to migrate both into one Flutter code base, and brought a project estimated at 6 to 8 months in at roughly 3 to 4, with the live apps unaffected.
- native apps into one codebase
- 2
- downtime
- 0
- feature parity
- 100%
Scooter’s Coffee’s two native code bases, iOS in Swift and SwiftUI and Android in Kotlin and Jetpack Compose, listing the same seven features. One at a time, agents read a feature in both, merge the two into one, write a Flutter spec, and scaffold it across Figma, Jira, and Flutter, and the feature’s screen lands in one Flutter app. Both native apps stay live throughout. The feature buttons send a feature through next.
iOS Swift · SwiftUI
Live
01The ask
Move two live revenue apps to Flutter and let nobody notice.
Scooter’s Coffee runs two mature, revenue-generating native mobile apps. iOS is built in Swift and SwiftUI, Android in Kotlin and Jetpack Compose, and together they drive hundreds of millions of dollars in annual revenue. When Scooter’s decided to unify on one shared Flutter code base, VGV joined the team to lead the migration with an AI-first approach.
This wasn’t a greenfield build. Customers depended on these apps every day, and they were the primary revenue driver for the business. Everything had to carry over, with full feature parity, the existing test suites, every backend integration, and a switch users would never notice, with no re-login and no downtime.
02The plan
Two workstreams, one platform.
VGV ran the engagement as two connected workstreams. The first was the native migration itself. The second was the Flutter development and product infrastructure that would carry the platform forward.
03Workstream 1
Three agents in a row, from native code to Flutter scaffold.
Instead of reading two mature code bases by hand and writing specs, VGV built a custom multi-agent pipeline that works as a cascade. Platform agents for iOS and Android analyze each code base, from features, architecture, and dependencies to test suites, integration tests, and documentation, and break the app into feature sets, functional areas, and requirements. A second agent turns that analysis into a Flutter application spec built against VGV’s published engineering standards and open source architecture repositories.
A third agent works across Figma, Jira, and Flutter to generate the components and scaffolding for the new app. It isn’t free, and it’s far faster than doing the same work by hand. The scaffold follows Feature-First Clean Architecture, the pattern VGV published. Our post Feature-First Clean Architecture in Production covers how it held up at 16 features and 88 packages.
04Workstream 2
Design a screen, get the widget and the ticket.
Alongside the migration, VGV brought AI-first workflows to the whole product team, so engineers, designers, and PMs work with AI tooling in their day-to-day. VGV built custom Claude Code skills that take Figma designs and scaffold Flutter widgets directly. A parallel set of skills reads the same designs and writes product backlog stories and requirements, so specs come from what was actually designed instead of being interpreted after the fact.
Design produces a screen. AI generates both the widget and the ticket. Engineering builds from specs that match the design exactly, and the gaps that usually build up between disciplines don’t get the chance.
05In the app
What a regular carries in their pocket.
-
-
-
-
- Plus the menu, drink customization, a store locator, and the cart.
Plus the menu, drink customization, a store locator, and the cart.
06What changed
Half the time, and the live apps never blinked.
-
Half the time
3–4 months against a 6–8 month estimate.
The multi-agent pipeline took the migration from native code bases to a standards-compliant Flutter scaffold in roughly half the original estimate.
-
Full parity
Features and tests carried over.
Feature and test coverage held throughout, and the live apps kept running unaffected.
-
One loop
Product, design, and engineering in one AI-assisted cycle.
Figma designs turn into both widgets and tickets, so engineering builds from specs that match the design.
-
One code base
Two native apps become one Flutter app.
Scooter’s Coffee will run on a single shared Flutter code base for iOS and Android.
07If you’re next
What we’d tell a team in Scooter’s seat.
- 01
Let agents read the native code first. A spec built from the real features, tests, and docs carries over more than one written from memory.
- 02
Set the parity bar before the first Flutter commit. Features, tests, integrations, and sign-in all come across, or the migration isn’t done.
- 03
Pick the architecture up front. Agents scaffold faster when there’s a published standard to build against.
- 04
Generate the ticket from the same design file as the widget, so the spec can’t drift from the screen.
Roll credits.
Who did what, and with which tools.
- Client
- Scooter’s Coffee
- Our part
- Migration leadMulti-agent migration pipelineFlutter developmentProduct infrastructureAI-first team workflows
- Platforms
- iOS and Android
- Built with
- FlutterClaude CodeFigmaJira
- Timeline
- 3–4 months, against a 6–8 month estimate
Built by Very Good Ventures
More success stories
All success stories
Scooter’s mobile ordering, moved to Olo in 5 days
- from 25% of Android users to everyone
- 5 days
- app uptime
- 99.98%
Divine: from vibe-coded to production-ready
- app stores at launch
- 3
- issues resolved
- 500+
A Braze optimization roadmap for Blade
An audit of Blade’s Braze setup, turned into a lifecycle roadmap.
We’ll care about your success as much as you do.
Our unicorns read every message and reply, usually the same day.
Start the conversationA 30-minute call about what you’re building.




