Devtrox is a small UK product studio. We publish our own mobile apps and build LaunchCrew, a crew of AI agents that takes indie apps from build to store to growth.
We work closely with the people who'll own the product after launch. No throw-overs, no deliverables for their own sake, just careful, opinionated work from the first sketch to the store listing.
Explore our servicesWeb platforms and internal tools built to outlive the trend cycle.
iOS and Android products that earn their permanent slot on the home screen.
Quiet, boring, predictable infrastructure. The kind we like.
AI agents and LLM features wired into the messy reality of your workflows.
Indie developers do the work of a whole team: compliance, store review, listings, marketing and promotion. LaunchCrew gives each of those jobs to a specialised AI agent, all working from your actual build. It grew out of everything we had to do by hand to publish Arrow Go and Mirror Beam.
See LaunchCrew in our workScans the build, metadata and screenshots for anything that would get the app rejected on the App Store or in Google Play Console, including policy breaches, missing permission reasons, misleading screenshots and accessibility gaps. It reports each issue with the guideline it breaks and a fix.
Maps every SDK in the build to the data it collects, then generates the privacy policy, Play Data Safety answers and App Store privacy labels. It covers country-specific rules such as GDPR and UK GDPR, CCPA/CPRA, COPPA, Brazil's LGPD and India's DPDP Act, and flags drift when a new SDK ships.
Writes the store title, subtitle, description and keywords, localises them into 20+ languages within each store's character limits, and drafts release notes from commits. It also produces A/B variants for every market.
Turns screen recordings and screenshots into marketing content: short vertical videos with cuts, captions and crops, store screenshots, feature graphics and post copy for each platform.
Connect TikTok, Instagram, YouTube, Facebook, X, LinkedIn and more once. AutoPostKit adapts each piece of content to every platform's format, schedules a posting calendar to promote your apps, and publishes after you approve.
We build a test set and a scorecard before the first prompt is written.
Every output passes a solver, a schema or a validator, wired in through tool use.
Model routing, prompt caching and batch jobs, so running costs stay predictable.
Personal data stays out of prompts. The same rule our apps follow.
Tell us about your team, the problem you're holding, and we'll come back with a one-page sketch of how we'd approach it — usually within 48 hours.