Case study
OnSense AI: AI Agents Across Web and Mobile
OnSense AI needed AI agents that were a working part of technician support and diagnostics, not a bolt-on feature, across web dashboards, backend APIs, and a mobile app. We built the platform's monorepo, the NestJS backend, a React Native app with an OCR nameplate scanner, and Appliance GPT, a standalone consumer AI chat product.
Industrial equipment repair · OnSense AI builds dedicated ML infrastructure for equipment repair: private LLMs and AI agents that help technicians, OEMs, and support teams diagnose and fix issues faster.
The brief
Move AI from a bolt-on feature to agents that are a working part of support, diagnostics, and technician workflows, across web dashboards, backend APIs, and a mobile app.
What was built
- Turborepo monorepo architecture for the platform
- Next.js web dashboards and NestJS backend APIs, including backend work for an ERP system
- A React Native (Expo) app for iOS and Android, including an OCR nameplate scanner that identifies an appliance's model from a camera photo instead of a worn or tiny model number typed by hand
- Appliance GPT (appliancegpt.onsense.ai), a standalone consumer AI chat product built solo in React and Next.js, so anyone can ask about error codes, parts, or repairs from a browser
- Model lookup with specs and part breakdowns, an interactive exploded-view parts diagram, an in-app GPT chat for repair questions, and parts cross-referencing against compatibility data
The hard part
AI agents are easy to demo and hard to make load-bearing. The brief here wasn’t “add a chatbot” — it was making agents a working part of how technicians actually diagnose and fix equipment, which meant the AI had to sit inside real workflows across a web dashboard, backend APIs, and a mobile app, not next to them. On top of the main platform, Appliance GPT was built as a separate, standalone consumer product solo, which meant carrying a second surface to production quality without the rest of the team to lean on for that piece.
Outcome
A production platform where AI agents are part of the support and diagnostics experience, plus a live, browser-based AI assistant for consumers. Both are running today, not sitting in a demo environment waiting for the next round of funding to justify finishing them.
Credit
Delivered by Abdullah Amir as part of a 2–3 person engineering team, before founding Bridge Innovate.