→
AI-first mobile app for identifying flies and insects from photos
What
What is
HatchMatch?
→
Transforms angler uncertainty into visible confidence matching
→
Built end-to-end as a solo designer / developer
→
The real design problem wasn't the camera-to-result flow — it was what the interface does when the model isn't sure
Why
Why
Agents?
→
Visual ambiguity requires reasoning, not rules
→
Identification requires judgment, not just pattern matching
→
Users need to understand why a result was suggested — not just what it is
→
A bad rig costs you fish. Uncertainty should be surfaced, not suppressed.
How
How I Used
Agents
→
The model makes classification and confidence calls
→
The app adds structure, guardrails, and context
→
Prompts were tuned iteratively based on real failure cases — not lab conditions
→
The confidence score is a design output, not just a model artifact — it shapes every UI decision downstream
→
When the model is uncertain, the interface says so — with a different rig recommendation posture and a clear visual treatment
Stack
Built With
What?
React Native
Cross-platform mobile
iOS beta-active
iOS beta-active
OpenAI Vision
Insect classification
Confidence scoring
Confidence scoring
Node.js
API layer
Prompt orchestration
Prompt orchestration
Supabase
Backend & auth
Session data
Session data
Vercel
Web hosting
Fast deploys
Fast deploys
Solo Build
Design + development
0 to live beta
0 to live beta
Status
Where It
Stands
Live Beta · TestFlight
Beta Testing — iOS Active
HatchMatch is in active beta testing on iOS via TestFlight.
Real anglers, real rivers, real feedback. The app is being refined
in the field based on actual use — not simulated edge cases.
If you fly fish, your feedback is worth more than a lab test.
Join Beta on TestFlight