What is an AI front desk?

An AI front desk is software that answers a business's customers 24/7 in place of a person at the desk. It answers their questions from facts the owner verified, flags what's urgent, quotes only prices the owner approved, and captures the job, on the phone, by text, or in a website chat, depending on the product. You'll also see it called front desk AI, an AI front desk assistant, or an AI front desk agent. Same idea, different labels.

That's the definition we'd defend. The term is young, so the rest of this post covers what the job involves, what it costs, and how to tell a real front desk from a chat window with a logo on it. One disclosure first: ServiceFront sells an AI front desk for plumbers, HVAC, electricians, and contractors. Ours works on the web and doesn't answer the phone, so weigh our definition the way you'd weigh any vendor's.

What does an AI front desk do?

Four jobs, in the order customers need them.

1. Answer the first questions. Do you cover my ZIP. Are you open Saturday. Are you licensed and insured. Do you do tankless. These are fact questions, and most of a front desk's day is this. The only acceptable source is a set of facts the owner has confirmed. A generic chatbot answers them from whatever it can guess, which is how a bot ends up promising Saturday hours you don't keep.

2. Triage. "There's water coming through the ceiling under the upstairs bathroom." "The furnace is blowing cold air and it's 20 degrees out." A front desk that knows the trade treats those as urgent and pushes toward a call right now; no hot water since yesterday gets a callback in the morning. Triage is trade knowledge, and it's the part a scripted operator can't do from an FAQ sheet.

3. Quote inside the owner's limits. The owner decides which services get a range, which get "that needs eyes on site," and which get nothing. The front desk never invents a number. If a vendor's demo produces a price you never gave it, that's the whole test failed; more on that below.

4. Capture or book the job. Name, address, the problem in the customer's own words, how urgent, the best time to reach them, delivered to the owner. With a calendar connected, the front desk books the slot; without one, it sets up the callback. (ServiceFront captures the lead and sets up the callback today. Calendar booking is on the roadmap.) Either way the conversation ends with a job the owner can act on, not a "thanks, someone will be in touch."

Booking is the last of the four on purpose. A front desk that opens with "want to schedule?" before it has answered anything is a scheduler with a chat skin.

One label, five products

Search the term and you'll find it on products that have almost nothing in common:

What's sold as an "AI front desk"	Where it answers	Who it's built for
An AI receptionist: a voice agent	Your phone line, sometimes texts and a website chat	Any business with a phone
A patient-intake suite	Calls, texts, and forms inside the practice's software	Clinics and dental offices
A check-in kiosk	A screen in the lobby	Offices, hotels, airports
A virtual receptionist	Your phone line, answered by AI with a remote person behind it, like Smith.ai	Businesses that want a human on the hard calls
A web front desk	Your website, a hosted page, and the links you text	ServiceFront's is built for the trades

They share one job. The channel is a detail. A front desk is the first person a customer talks to: the one who knows the hours, the service area, whether you're licensed, roughly what a water heater swap runs, and whether a sewage smell in the basement can wait until Monday. Phone, chat, or a lobby screen, the facts and whether they're right are the product.

ServiceFront is the last row. If your problem is the phone, the spec-level comparison with AI receptionists and human answering services is the honest place to start.

How is an AI front desk different from an AI receptionist?

An AI receptionist is one channel of the job: a voice agent that answers your phone line. An AI front desk is the whole job, from the first question to the captured lead, on whatever channel the customer uses. Some products do both. Here's how the front desk compares with its neighbors.

An AI receptionist. A close cousin on a different channel. Plenty of good ones exist. If the phone is where you lose customers, that's the category to shop.

An answering service. A human operator at a call center, working from your script, billed by the minute or the call. The strength is a human voice. The ceiling is that a script can't triage.

A website chatbot, 2019 edition. The button tree that ends in "leave your email and we'll get back to you." That's a contact form with extra steps, and the contact form is where leads go to die.

A scheduler. Calendar booking is one outcome of a conversation. A front desk that only books is answering the easy half of the job.

What does it answer from?

Every AI front desk is a machine for answering with facts, so the only question that matters is where the facts come from. Vendors give three answers:

• "A prompt we write during onboarding." Weakest. Someone typed a paragraph about your business once, and the agent improvises the rest.
• "We crawl your site." Better, but your site is probably out of date, and a crawl can't tell your real Saturday hours from the ones you posted in 2022.
• "A profile of facts you verified." The right answer. The owner reviews what the agent believes before it talks to a customer, and anything outside that profile gets "I'll check" plus a captured lead, never a guess.

This matters because what your front desk says, you said. In February 2024 a British Columbia tribunal held Air Canada liable for a bereavement-fare answer its website chatbot gave that contradicted the airline's actual policy, and it rejected the suggestion that the chatbot was a separate legal entity responsible for its own actions (Moffatt v. Air Canada, 2024 BCCRT 149). The damages were a few hundred dollars. The rule it set is the expensive part: a made-up answer is a promise your business has to keep or clean up.

What happens when it can't answer?

A good one says so. When a question falls outside the facts the owner verified, like a price nobody set or a town that isn't on the service-area list, it says someone will confirm, takes the customer's number and the problem in their words, and hands the whole thing to the owner as a lead. That handoff is the part to test. A front desk that guesses to keep the chat going is worse than no answer at all, because the guess becomes a promise. ServiceFront's front desk also marks emergencies, like a burst pipe or a gas smell, as urgent in the lead it emails you.

Where does an AI front desk work?

Wherever the customer shows up. For ServiceFront today that's the web: your website, your hosted page at yourbusiness.servicefront.ai, the link on your Google profile, and the link you text back when you can't pick up. The customer it covers is the one phone coverage never touches: the homeowner comparing three plumbers at 11pm who isn't going to call anyone tonight, and who may ask ChatGPT instead of reading your site at all. Answering inside those assistants is where the category is headed. For ServiceFront that's on the roadmap, not shipped.

How much does an AI front desk cost?

It depends on the channel, and the pricing shape matters more than the sticker price. Published plans we checked on September 28, 2026:

• AI phone receptionists. Rosie runs $49 a month for 250 minutes up to $299 for 2,000. Goodcall runs $79 to $249 a month per agent with unlimited minutes, plus $0.50 for each unique caller past the plan's 100 to 500 a month.
• Human answering services. They bill for talk time. Ruby goes from $250 a month for 50 minutes to $1,725 for 500.
• ServiceFront. A free tier that answers on your site and emails you your first 5 leads a month, and emergencies even past that. Basic is $59 a month flat for every lead.

Flat monthly is the right shape for a front desk. Per-minute and per-call meters mean your busiest month is your most expensive month, which is backwards. A front desk should cost the same in a slammed August as in a slow February. The comparison post has more published numbers for the phone side.

Is an AI front desk worth it for a small shop?

It's worth it when the jobs it catches are worth more than it costs. Angi puts a typical plumber bill at about $340, so one saved job pays for close to six months of a $59 plan. The jobs a web front desk catches are the ones you'd otherwise never see: the 11pm website visitor who was never going to call, and the caller you text a link to while you're under a sink. If a dispatcher already answers everything within minutes, day and night, it's a smaller win, and we've written down who we're not for.

Will an AI front desk replace a receptionist?

In a solo shop there's no receptionist to replace. The owner is the front desk, and the AI covers the hours when nobody can answer. In a shop with a CSR, it takes the repetitive questions and the after-hours traffic, and the CSR keeps the conversations that need judgment: an angry customer, a warranty dispute, a job that doesn't fit the price list.

Who is it for?

Service businesses where the owner is the front desk: the solo plumber, the two-truck HVAC shop, the electrician whose office is a phone in a pocket under a sink. If you already have a dispatcher and a CSR team, you have a front desk, and software that answers the web is a smaller win for you.

How do you test one before you trust it?

Put three questions to any vendor's demo, ours included:

1. Ask for a price you never gave it. The right answer is that it won't produce one.
2. Ask "am I talking to a person?" The only acceptable answer is an honest one. Here's why that matters more than it seems.
3. Ask something it can't know, and watch what it does. Acceptable: it says so and takes a message. Unacceptable: a confident guess.

The full seven-question checklist is here, with a scorecard of where ServiceFront passes and where it fails. To run the test on us, paste your site's URL at servicefront.ai and ask the front desk the questions your customers ask you.