For 3+ years, Marty and I have been building Tabbi AI. Real users, real revenue. Enough to cover costs, never enough to pay ourselves properly. The kind of trajectory where you're not failing, but you're not going anywhere fast either. Passion was leaking.
Somewhere in that stretch, Marty drifted into a side project: video chat infrastructure. No target user, no business model. Just an engineer following what was interesting. He'd still come back once a week for Tabbi conversations, but his energy was clearly elsewhere.
Meanwhile, I'd gone deep into vibecoding.
In a few weeks I shipped six apps I'd been carrying in my head for years: a todo app, a meal planner, a childcare-duty tracker for our family, a new version of TabbiMate for language learners. For the first time in my career, the bottleneck wasn't my coding skill. It was my imagination.
The other thing I noticed: I was spending money on software tools faster than I ever had. Cursor, Claude, extra tokens, every new editor add-on. More in one year than I'd spent on tech tools in my entire life before. Not because I had to. Because vibecoding is fun, and every tool that made it more fun felt worth it instantly.
In one of those weekly Tabbi conversations, I pitched a pivot: connect language learners over video chat, using what Marty was already building. Then, half-joking, I said:
“Why don't I just vibecode the whole thing, and you give me video chat support?”
That was the moment. Not because of TabbiMate. Because of what it implied.
If vibecoders had drop-in video chat, a whole class of products becomes possible that currently isn't. Coaching apps. Tutoring tools. Group hangouts for niche communities. Customer support widgets. Language exchanges. Half the ideas I'd been sitting on could suddenly be built by one person in a weekend.
The existing tools weren't going to get us there. Too complex. Every “just add video” library assumes you already understand WebRTC, ICE candidates, SFUs, TURN servers. Too expensive. Pricing built for venture-backed startups, not indie devs paying out of pocket. And critically: none of them were built for how vibecoders actually work, with an AI assistant in the loop.
Marty had been quietly building the answer for months. He just hadn't met the people who needed it yet.
So we built VibeLive.
The idea, in one line
Drop in video chat the way you drop in Stripe. Nobody builds Stripe to add payments. They use Stripe. There hasn't been a Stripe for video. That's the gap.
What makes it vibecoder-native
An AI guide file. Drop it into your project and Cursor, Claude Code, or whatever you're vibecoding with knows the entire SDK: every method, every option, every gotcha. No more pasting docs into chat and hoping. Your AI assistant just knows how to use it. That's the piece none of the existing players have.
Who we are
Marty is the engineer and CEO of Makedo Inc. I run design, product, and marketing. Tabbi AI keeps running. VibeLive is our new bet.
We're bootstrapped. Free forever for hobby projects, paid plans when you need more. We're building for people like us, and we're not going anywhere.
If you've been vibecoding and want to build something with video in it, try VibeLive. Marty and I are in Discord daily. Come tell us what you build.
— April