I ship voice AI that does a job, not chat demos.
A customer can request a plumber and get a real callback. A business can paste its website and go live with a receptionist. A job-seeker can practice an interview out loud. That is the work in this profile.
Start here: oscaller.com (live) · GPT-CV demo
| Project | What it is | Try it |
|---|---|---|
| OSCaller | Home-service dispatch. Intake → nearest available tech → Aria (Retell) calls to confirm. Maps, OTP, Stripe. | oscaller.com |
| meclients | Done-for-you AI front desk. URL in, voice agent + phone number out. Stripe + Twilio + Retell. | repo |
| GPT-CV | Live A4 CV studio, cover letters, ATS check, and a spoken Interview Lab. | demo |
| Seedance Studio | Desktop app for Seedance 2.5 video generation (Electron, not Vercel). | npm run dev |
| PRISE Sarl | Civil-engineering brand site. Ernest, a FR/EN realtime agent, can tour the company and take a lead. | npm run dev |
Every repo has a README, a placeholder .env.example, and no live secrets in the tree.
Next.js · TypeScript · Supabase · Twilio · Retell · OpenAI Realtime · Stripe · Google Maps · Vercel · Electron
Voice, maps, payments, and auth are not bolt-ons. They are the product.
- Click OSCaller and GPT-CV — they are deployed.
- Skim one README (they all tell you stack + how to run).
- If you clone:
copy .env.example .env.local— keys stay on your machine.
Happy to walk through any of these on a call.