Picture a storm rolling through your town on a Tuesday night. By 9:30, a few dozen homeowners have asked their phone assistants the same thing: find a roofer who can inspect hail damage this week and tell me roughly what it costs. Each assistant contacts several roofing companies. By morning, your phone, inbox and web form are holding what amounts to fifty inquiries that all arrived at once.
Fifty is an example number. The point is the shape. Many requests, the same hour, the same questions, and your competitors got them too.
One request, many businesses
When a person shops for a service, she contacts businesses one at a time. When she hands the job to an assistant, the assistant can contact many at once, because extra calls cost it nothing. One request fans out across a whole category of local businesses. And it happens whenever the customer thinks of it, which is often after dinner, not during your office hours.
This is no longer theoretical. Google already offers AI calling to local businesses for U.S. Search users, with each call opening by saying it is an automated system calling on a user's behalf. A newer Gemini feature in testing places calls from the user's own phone number, so you will not always know which callers are assistants. For the background, see how customers started delegating the legwork.
The work is small. The timing is the problem.
Here is what surprises owners. Each of those inquiries is easy. A ballpark range, a yes on the service area, a couple of open times. Your office manager could answer any one of them in two minutes. The trouble is that a person answers one at a time, and the inquiries do not arrive one at a time. They arrive in a burst, after hours, and they expect a reply before the customer wakes up.
So the constraint is not effort. It is simultaneity. The business that can answer fifty inquiries in the same hour, with the same quality as one, gets the shot at the job. That is a clockspeed question: how fast a request becomes a useful answer, at any volume.
The seven questions you will hear
AI-generated inquiries tend to ask the same things. Prepare an approved answer for each one, written down, so nobody improvises at midnight.
- What's the ballpark price? Write a range or a rule you are comfortable with, like a starting price for an inspection or a typical range for a common job. Never a made-up quote. If price depends on a site visit, say that and offer the visit.
- Do you service my area? List the cities, ZIP codes or radius you cover, plus what happens just outside it: a travel fee, a referral, or a polite no.
- What's your availability? This should come from your real calendar, not a guess. If the answer is "next Thursday," the agent should be able to see that.
- Do you offer financing? Yes or no, the name of the program if there is one, and who handles the application. Keep terms out unless you have approved the exact wording.
- Can you schedule me? Decide which appointments can be booked directly and which need a person to confirm first.
- Are you licensed and insured? A short, accurate statement you can stand behind, plus where to verify it.
- Can you send more info? A ready text or email with your services, service area, how pricing works and a booking link.
For most first contacts, that list is the whole job.
Why voicemail and slow callbacks fail
An assistant reporting back to its user has a simple story to tell: who answered, what they said, and who could book. Voicemail fits under "did not answer." A callback the next morning arrives after the customer has already read a summary of three businesses that did answer. You may still reach her, but now you are calling to compete with a decision she has half made.
Answering services help with pickup, but a message is still a delay.
Traditional
A night of messages, a morning of phone tag
- Burstarrives
- Voicemailfills
- Officesorts
- Callbacksone by one
- Phone tagrepeat
- Few bookedlate
AI native
Every inquiry answered the same night
- Burstarrives
- AI agentanswers all
- Calendaroffers slots
- Text, emailfollow up
- Bookeddone
Where people stay: large jobs, complaints and anything outside the rules go to a person with a summary.
How an agent handles the burst
A voice agent can take many calls at the same time, and it answers an assistant the same way it answers a person. It gives your approved answers to the seven questions, checks your calendar, and offers a time. No special AI-to-AI protocol is involved, just clear answers and access to your schedule.
Then the follow-up does the rest. Right after the call, the customer gets a text with a booking link and an email with the information packet you approved. Written details are easy for an assistant to pass along. Inquiries from your web form or email get the same treatment. For the qualifying side, our lead qualification agents sort the serious inquiries from the tire kickers using your criteria.
Every inquiry gets the same answer, however many arrive, and lands in your CRM.
Where humans take over
Volume is the agent's job. Judgment is yours. The agent hands off when a conversation turns high value, unusual or sensitive: the commercial property manager with twelve buildings, the homeowner who is upset about a previous contractor, the insurance claim that needs a careful explanation. Your team gets those with notes attached, during hours you choose, and can spend real time on them because the routine fifty are already handled.
What to measure
Watch response time on after-hours inquiries specifically. Track how many inquiries get a complete answer to the seven questions on first contact, and how many reach a booking without a callback. Count handoffs per booked job. Compare capacity before and after: the hours your office spends on repeat questions versus customers who need a person. Those numbers tell you whether your business is ready for assistant-driven inquiries or still running at the clockspeed of a voicemail box.
The AI Audit is a paid two-week engagement, credited toward your first build. We review how your business handles a burst of inquiries, after hours and on every channel, and show you where they stall.
Get an agent readiness check → What agent-ready means →Questions owners ask
Will the agent give out prices I have not approved?
No. The agent only uses the ranges and rules you write down and approve. If a job needs a site visit or a custom quote, it says so and offers to book the visit. Anything outside your approved answers goes to a person instead of getting a guess.
What if a real person calls during the same burst?
They get the same treatment. The agent answers people and assistants the same way, with the same approved answers and the same booking options. Callers who want a human, or whose request falls outside your rules, are handed to your team with a summary of the conversation.
Do I need to answer all seven questions before we start?
No, but you need most of them. Service area, availability and how to book are the minimum. Price ranges and financing can follow. Part of the audit is drafting those answers with you, so they are accurate and in your own words.
Related: The Fastest Response Wins: How AI Assistants Pick a Business · Why Speed to Lead Decides Who Wins the Job