Day 4: The Voice Interface

Typing on your phone works fine at a desk. It doesn't work when your hands are covered in thinset or you're holding a level against a wall.

So I built a voice interface. Talk to Jarvis, it talks back. Think Iron Man, minus the suit.

How It Works

• ElevenLabs for the voice — natural-sounding text-to-speech
• Web Speech API for listening — runs in any browser
• Simple web page on my local network — open it on any device

Now I can be on a job site, hands full, and say: "Jarvis, what's the lead time on that vanity for the McCartney job?" And get an answer out loud.

Or: "Jarvis, text the client that we're running 30 minutes behind." Done.

Or: "Jarvis, what's on my calendar tomorrow?" Instant rundown.

Why This Matters for Contractors

We don't work at desks. Our "office" is a job site, a truck cab, a lumber yard. Any AI system that requires you to sit down and type is built for office workers, not us.

Voice changes the game. It means AI fits into your actual workflow — not the other way around.

Total build time: about 3 hours. Cost: ~$22/month for the voice service. That's less than your Spotify subscription.