Night Rider – AI Researcher App Development

A year ago, I designed and built my most ambitious side project to date: Night Rider, an AI researcher app.

The pitch was simple: You sleep. AI works. Wake up to a custom audio briefing.

You submit a question or task before bed, anything from “How does gradient descent actually work?” to “Summarize my inbox”, and the app processes it overnight. In the morning you get a 3–5 minute audio recap to listen to on your commute, during a workout, or over coffee.

I’m a big podcast listener, but I kept wishing I could wake up to something made for me — not a generic feed, but a briefing on whatever I was actually curious about that week. Night Rider was my attempt to build that.

This was end-to-end product work, not just a prototype:

✷ Product design: brand identity, UX flows, app UI, and the marketing site.

✷ Mobile app: shipped on iOS and Android.

✷ AI pipeline: overnight background processing with LLMs (OpenAI, Perplexity), text-to-speech (ElevenLabs), speech-to-text (Deepgram), and the AI agent running on LiveKit.

✷ Monetization & Infra: subscriptions via RevenueCat, analytics via PostHog, data storage via Supabase, background jobs via Upstash.

✷ The landing page: Astro + React + Tailwind, deployed on Vercel.

Running a solo AI product taught me more about real-world product constraints than any client project. API costs add up fast. Overnight processing needs careful orchestration. And “cool AI demo” and “sustainable business” are two very different problems to solve.

I sunset Night Rider this year 💔 It wasn’t sustainable to keep running, and I’d rather be honest about that than pretend it’s still live. But I’m proud of what it became — a full product, in the App Store, that actually worked.

If you need someone who can design and build AI-powered products from scratch, this is the kind of work I do.