BytePilot came out of running AI agents on real business phone lines — taking real bookings, from real customers, with real consequences when a call went wrong. Months of that taught us the things you only learn the hard way: how long a pause can be before a caller says "hello?", why an agent must never guess a date, and what it costs when one hangs up on somebody who hadn't finished speaking.
Those lessons are in the engine, not in a roadmap. The templates we ship are the ones we've run ourselves, and every new one has to pass a scripted test suite of awkward calls before it goes anywhere near a customer's phone line.
Voice came first because it's the hardest — a conversation in real time, where a half-second pause is noticeable and a wrong date costs someone a booking. Web chat and direct model access follow on the same platform, the same balance and the same white-label rules. What we're building is the infrastructure layer for people who sell AI services, not a single product.
The businesses that need this most — the dental practice, the plumber, the restaurant — don't buy AI platforms. They buy from someone they already trust, who set up their website and answers the phone when something breaks. That person deserves the margin, and their client deserves to keep dealing with them. So we sit behind you and stay out of it.
You pay per minute, we publish the rate, and we don't sell to your clients. That's the whole model. No seat licences, no annual commitments, and no surprise tier that appears once you're dependent on us.
Built and run in the UK by Apexia Group Limited. Data stays in the UK and EU, and you can reach an actual person at support@bytepilot.ai.