How to use Unknown to build an air quality app
The author reports App Store approval exactly 60 days after starting an empty repository, with a global PM2.5 map, charts and sensor integration. He says native iOS and Android apps were live.
news.ycombinator.comPrompt to try
Editorial adaptation, not the source author’s exact prompt. Replace bracketed inputs and review tool permissions first.
Use [existing web app], [screenshots] and [Figma mocks] to draft a native [platform] app specification covering APIs, data models, localisations, UI and error handling. Implement [feature] from the approved specification and identify decisions and debugging work that need my input.
Reach people exploring AI agents
See placements →What happened
The author reports App Store approval exactly 60 days after starting an empty repository, with a global PM2.5 map, charts and sensor integration. He says native iOS and Android apps were live.
Reported by ahaucnx on Hacker NewsHow the agent did it
- Asked the AI to draft full specs from the existing web app.
- Supplied screenshots and Figma mockups to guide the specs.
- Let the coding agent generate code while handling company work.
Where a person stepped in
The author explicitly describes debugging, architectural decisions and understanding difficult implementation details.
A person approved each step before the agent continued.
AirGradient's CEO used an unnamed coding agent to build the company's first native iOS app while continuing his leadership work. He had coded on the side for about 15 years but had never used Swift or SwiftUI. He reports moving from an empty repository to App Store approval in exactly 60 days, working on the project on the side.
The app presents a global PM2.5 map with detail views, charts and integration with AirGradient's open-source sensors. To guide implementation, he asked the AI to draft full specifications from the existing web app and supplied screenshots and Figma mockups. Some specifications ran to several pages, covering APIs, data models, localisations, UI mockups and error handling, even for a simple feature.
While the agent generated code on one virtual desktop, he handled emails, tickets, proposals and strategic planning on another. He still debugged the app, made architectural decisions and worked through difficult parts. He says the app was live on iOS and Android, with a native Kotlin Android version, while identifying it as version one with improvements planned.
Who it's for: Technical leaders with some programming experience who can supervise implementation alongside other responsibilities.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Hacker News · ahaucnx
Frequently asked questions
What inputs guided the app implementation?
The author used the existing web app as the basis for full specifications and supplied screenshots and Figma mocks. Specs covered APIs, data models, localisations, UI and error handling.
Did the coding agent replace the author’s engineering work?
No. The author still debugged, made architectural decisions and understood the tricky parts, while the agent generated code alongside his CEO work.
More details
Remaining steps
- Debugged the app and made architectural decisions.
About this account
The coding agent generated implementation code while the author worked on company tasks on another virtual desktop.
Platform: Hacker News. Attribution: https://news.ycombinator.com/item?id=45922456. Prompt: Suggested (written by Agents In Use, not the author's). Sourced by the Agents In Use pipeline.
Credit: ahaucnx on Hacker News
Source published 14 Nov 2025
Source checked 14 Oct 2026
Tools and topics
Unnamed coding agent · Swift · SwiftUI · Figma
iOS · SwiftUI · Air quality
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.