What each one is, how it bills and what it connects to
In short
All three sell an AI receptionist as a monthly plan on their own platform, priced by calls, by callers or by minutes, and each page names ways for it to book and to reach other software. Ours is a build for one business, with a setup fee before launch and a care fee each month after it, while the phone and AI vendors bill each call to accounts you own.
Smith.ai prices its AI receptionist by the call. The free plan covers 25 calls a month, with a price for each call after that, and its card lists AI scheduling and native integrations, Zapier among them. Pro, which you set up yourself with its support, is a monthly plan of calls, and Enterprise, its largest plan, is one its own team builds and runs, the only plan its feature guide shows with custom integrations and custom AI prompting.
Goodcall bills by caller instead. Its Starter plan is a monthly price for 100 unique customers, with a rate for each one after, and its page says that talking longer or calling more often never changes the bill. You build the agent yourself from a free demo agent, and Starter includes five workers, which its FAQ says can be a booker or a message taker, and one external tool, meaning anything you connect or build, such as an API.
Of the plans on its cards, My AI Front Desk prices one, Business-in-a-Box, by the month or at a lower monthly price billed annually, with 200 voice minutes and a monthly pool of 1,000 overage credits that each minute past the 200 draws on. That plan carries web chat, two-way texts and a CRM of its own, the page says every plan connects to calendars, CRMs and other apps through Zapier, and its Partner and Enterprise plans, priced on volume, add API access and custom integrations.
| How it bills | What its first plan includes | What its page says it connects to | Who sets it up | |
|---|---|---|---|---|
| Smith.ai | By the call | A free plan of 25 calls a month, with a price for each call after | AI scheduling and native integrations, Zapier among them; custom integrations on Enterprise | You, with its support, or its own team on Enterprise |
| Goodcall | By unique caller each month, never by the minute | A monthly plan for 100 unique customers, with a rate for each one after | One external tool on Starter, such as an API you connect | You, starting from a free demo agent |
| My AI Front Desk | A plan of voice minutes, then credits | A monthly plan of 200 voice minutes, with a pool of overage credits | Calendars, CRMs and other apps through Zapier; API access on Partner and Enterprise | You, or its team on Partner and Enterprise plans |
| A custom one: ours | A setup fee, then care each month; the calls billed at cost by the phone and AI vendors | One line, one booking tool and one CRM, set in a written plan | One booking tool and one CRM, through each one's own API where the written plan confirms it, or by request and email | Us, from a written plan you approve |
Swipe sideways to see every column
Where each one fits better
In short
Smith.ai fits when you want its own people behind the AI at any hour, Goodcall when calls run long and you would rather build the agent yourself, and My AI Front Desk when you want chat and texts in the same plan. A custom one fits when you want the number and every account behind it in your own name, set up for you to a written plan.
The FAQ on Smith.ai's pricing page says a call too complex for its AI can be transferred to one of its live receptionists, based in North America, at any time of day or night. No staff of ours sits behind a receptionist we build: in your hours it puts calls through to the number you name, and after hours it follows your one rule, a message for the morning or urgent calls sent to your on-call number. If you want Smith.ai's own people to take a hard call at night, Smith.ai is the better fit.
Goodcall's per-caller billing suits a line where calls run long, since its page says talk time does not change the bill, and you can try a demo agent for free before you add a number. It also suits an owner who would rather build the agent and connect its tools than have it built.
My AI Front Desk puts web chat, texts and a CRM of its own in one monthly plan, which suits a business that has none of those yet, and its page lists 20+ languages, where ours answers in English. Its page also says setup takes minutes.
Reaching your tools is not, by itself, a reason to build, since every one of these pages lists a way to do it, and Smith.ai's Enterprise plan is built and run for you as well, on its own platform. A build of ours fits when you want the phone and AI accounts opened in your name, so the number and each call's log stay with you, and its hookup to your booking tool and CRM checked against each tool's own API in the written plan, ahead of the build. You pay more at the start for that, and you wait for the build and its test calls before it answers your callers.
What a custom one covers, and what it does not
In short
Setup covers a single line hooked to one booking tool and one CRM, and care each month keeps it running after launch, while the calls' minutes and AI usage bill at cost to your own accounts.
Where the written plan confirms that your booking tool's maker documents a way for software to add bookings, and that your account may use it, the receptionist books straight in; where it cannot, it takes a booking request your team confirms, and tells the caller which of the two happened. Each call reaches your CRM as a card on the same terms, or your inbox as an email.
Before it answers anyone, its script and its rules for each kind of call go into a plan you approve, and test calls run on its own number while you listen in or place your own. After launch, care keeps it in step with your hours and prices.
It answers in English and takes no payments. It never reads a caller's details back from your records, and texts sit outside its starting scope: texting your callers is quoted on its own and waits until a texting registration in your name is approved. We do not offer a BAA today, so we never build one where patients' health details would reach it.
Projects start at $2,500. Book a call, and the written plan that follows prices the setup and the monthly care for your line before anything is built.
A worked example (hypothetical)
In short
A locksmith who works nights alone could pick any of the four, and the choice turns on who should take a call the AI cannot settle and whose accounts the line should sit on.
Picture a locksmith in Koreatown who works nights alone and keeps every job in a scheduling app on the van's tablet. When the phone rings in the middle of a job, nobody is free to pick up.
Each plan could answer those calls, and each lists a way to book. Smith.ai's live receptionists could take over when its AI cannot settle a call, Goodcall's bill would stay the same however long a locked-out caller talks, and My AI Front Desk could reach the scheduling app through Zapier if that app is one Zapier connects to.
A build costs more on day one, with its setup and its monthly care priced in a written plan, and it fits if the owner wants the line and its records on accounts the shop owns. Its written plan would first check the scheduling app's own API: with a way in, the receptionist books visits straight into it; without one, each call reaches the owner as an email with the request in it. Either way, an urgent caller goes through to the owner's phone.
What this means for a small business
In short
Each of the three lets you try it before you pay, so a test on your own calls costs little. Price a build when you want the number and its accounts in your own name, and the hookup to your tools confirmed in writing first.
Before a trial, list by name the tools a call has to reach, and look for each one in the company's own list of connections, or ask its sales team. Ask as well what happens to your number, and to the history of past calls, if you leave the plan.
We read the three pages on October 16, 2026. Plans and prices change, so read each page again on the day you choose.