ProductProduct & AI Engineer2026in-progress
Stride
An AI-first adaptive wellness platform that turns tracking into a behavioral companion.
Stack
Next.jsConvexTailwind CSSOpenRouterVercel
Outcomes
Multi-modalText · voice · image · barcode
5 agentsDiet · workout · sleep · hydration · habits
AdaptivePersonalised over time
What it is
An AI-first wellness platform built around one principle: tracking should feel effortless. Stride combines multi-modal input, agentic AI, and a behavioral memory layer so the system adapts to the user instead of forcing the user into rigid forms.
Key points
- Multi-modal nutrition logging — text, image, barcode, and nutrition-label parsing flow into a deterministic calorie and macro calculator, with AI used for understanding rather than as a black box.
- Natural-language workout tracking — the system parses "30 min run, 3x10 squats at 60kg" into structured sets/reps automatically.
- Five specialised agents — diet, workout, sleep, hydration, and habits agents share a behavioral memory layer, making the platform feel like a coach instead of a spreadsheet.
- Behavioral memory — preferences, consistency patterns, interaction styles, and adherence are stored over time so recommendations get sharper the more the user engages.
- Adaptive UX — casual users get quick logging; advanced users get detailed analytics; overwhelmed users see simplified interfaces — all driven by behavioral signals.
- Modern stack — Next.js + Tailwind on Vercel, Convex for real-time state, OpenRouter for LLM routing across providers.
Result
Architected as a six-layer system: input, behavioral memory, intelligence, coaching, analytics, and adaptive optimisation. Currently in active development with the input and intelligence layers operational.