An AI receptionist for a wedding planner should mostly answer the inbox, not the phone

Count where your last 50 inquiries came from before you buy anything

Most products sold as an "AI receptionist" answer phone calls. Before you pay for one, check whether couples actually call you. It takes about 20 minutes.

1. Search your inbox for the subject line your website form sends, and count the results.
2. Search for notification emails from each wedding directory you list on, such as The Knot, WeddingWire or Zola.
3. Scroll back through your Instagram and Facebook message requests.
4. Open your phone's call log and count calls from unknown numbers that turned out to be couples.

Stop when you reach 50 first contacts. Then use this rule. If calls make up fewer than about one in ten, a phone bot is answering the wrong door. If calls are more than one in five and some go to voicemail and never come back, phone answering is worth paying for. If you land in between, a good voicemail greeting that points callers to your form and promises a same-day callback is usually enough.

What the receptionist should do, and which parts AI should never touch

A couple's first message almost always asks three things. Are you free on our date? Roughly what does it cost? What happens next? A good first reply answers all three. Only one of those answers needs AI.

Here is a worked example. A form comes in at 10pm on a Saturday: "We're marrying at Hollow Creek Farm on June 12, 2027, about 120 guests. Do you do month-of coordination?" A rule checks June 12, finds it open, and sends a short acknowledgment with your pricing guide attached. Separately, the AI drafts a fuller reply. It confirms the date is open and says month-of coordination is available. It asks whether they have a planning budget in mind, because they did not mention one, and it links your booking page. You read it on your phone Sunday morning, fix one line, and send.

Keep the personal reply on draft-and-approve. Do not let it send automatically. A confident wrong sentence about what your package includes costs more than a reply that goes out a few hours later. This split between plain rules and one smart step is the same pattern covered in AI automation is mostly plain rules with one or two smart steps.

What an inbox receptionist costs when you use your own API key

Drafting one reply means the model reads the inquiry, your package notes and your instructions. Call it about 1,500 tokens in. It then writes about 400 tokens out. On a mid-tier model such as Claude Sonnet, listed at $3 per million input tokens and $15 per million output tokens when we last checked, that works out like this:

• Input: 1,500 × $3 ÷ 1,000,000 = $0.0045
• Output: 400 × $15 ÷ 1,000,000 = $0.006
• Total: about $0.01 per drafted reply

Prices change, so check the current rate on Anthropic's pricing page or OpenAI's and redo the two multiplications. A busy planner getting 60 inquiries a month spends about 63 cents. Even 200 inquiries is about $2. That fits the range in what an average AI task costs. The way to blow it up is to paste your whole 40-page planning guide into every request. Send only the package summary instead.

This is the part Fynl is built for. A form on your site lands in the CRM, a rule checks your calendar, and the AI drafts the reply in your Gmail-connected inbox using your own Claude or ChatGPT key. The cent goes straight to Anthropic or OpenAI with no markup. It costs $49 a month for the whole portal. It does not answer phone calls or send texts, so if your tally said phone, read the next section instead.

When a phone AI receptionist is the right buy, and it is not us

Some planners do get calls. Couples ring from the car after a venue tour, parents phone instead of filling in forms, and you are at an event every Saturday with your phone in a drawer. If your tally showed more than one in five inquiries by phone and you are missing them, pay for phone answering.

You have two kinds to choose from. AI phone receptionists answer, take details and can book calls. Smith.ai offers both AI and human-backed answering. Human services such as Ruby cost more and handle emotional or unusual callers better. Neither publishes a price that will stay current, so compare on their own pricing pages and ask these four questions:

• Can it read your real calendar, or will it take a message about a date it cannot check?
• Is it billed per call, per minute or a flat monthly fee? Wedding calls run long.
• What does it say when a couple asks about price?
• Can it hand the call to your mobile when you are free?

Whichever you choose, the phone service and the inbox drafting solve different problems. Many planners need only the second.

Questions people also ask

Sources

• Anthropic API pricing
• OpenAI API pricing
• Smith.ai answering services
• Ruby receptionist services