The questions owners actually ask

Those route calls or take messages. This actually handles them. An answering service writes down “customer called about water heater” at 2 AM; the agent books the 8 AM slot, texts the confirmation, and logs the job — because it knows your services, your prices, and your schedule. Messages create work for the morning. Handled calls don’t.

It will occasionally — anyone answering a thousand calls a month does. Our job is to make mistakes rare, small, and caught. The agent only answers from knowledge you’ve approved, it’s built to say “let me get someone for you” instead of guessing, and anything that moves money or can’t be undone waits for a person to approve it. We review its conversations, especially in the first weeks, and fix issues at the source. What we’ll never do is promise perfection — we promise accountability.

Compare it to what missed calls already cost. If your average job is worth a few hundred dollars, a handful of missed after-hours calls a month is the fee — and that’s before counting your own time on repetitive phone work. For context on the alternatives: do-it-yourself AI platforms are cheaper but leave all the hard configuration to you, and fully custom builds typically run $50,000 and up. We sit in the middle: a working, managed agent for a fraction of a custom build, priced by how much you actually use it.

You can experiment, and some owners do. What they usually discover is that the chat part is the easy 20%. The hard 80% is phone lines, connecting your calendar and job software, deciding what the agent is allowed to do, teaching it your actual policies, and watching it so it stays right. That’s the service. You’re not paying for AI — you’re paying for AI that knows your business and someone accountable for it.

You own it — your customer list, call recordings, documents, and the agent’s configuration are yours. Nothing from your business is ever shared with another client, and the AI providers we build on state in their commercial terms that customer data is not used to train their models. If you leave, we help you export everything and then delete what we hold. The full picture is on our Your data page.

It says so, and hands off. The agent is deliberately built to escalate rather than improvise: unknown question, upset caller, anything outside its approved scope — it takes a message with full context or transfers to your team, your choice. A confident wrong answer is the one thing we design against hardest.

Typically a few weeks from kickoff, not months. We need a discovery conversation, access to the tools it should connect to (like your calendar or job software), and your sign-off on what it will know and what it’s allowed to do. Then it launches quietly — closely watched — before taking on full volume.

No long lock-in — either side can end the agreement with 30 days’ notice. If you leave, we help you export your knowledge, recordings, and records, then delete what we hold. We’d rather earn next year than lock you in — and we agree on success numbers in writing before launch, so you’ll know early whether it’s working.