None of the AIs booked live on the call

By BotsRanked Test Team · November 2026 · Round 1

By BotsRanked Test Team · November 2026 · Round 1

Across eight completed demo calls to six AI receptionist vendors, we could not verify a live booking. Getting a person involved meant a callback offer, more intake, a separate phone, email or chat contact, or a transfer the business would have to set up. My AI Front Desk offered a text booking link, and Dialzara offered a booking link. Neither AI-Receptionist.com demo promised a callback; both said no human would follow up.

Every call reached a vendor demo line, not a paid plan. We did not configure these systems with the same business facts or connect them to a calendar we could inspect. Our caller also asked that nothing real be booked and declined to provide contact details. These calls show how the demos responded to that script. They do not establish whether a paid setup can book appointments.

Scorecard

  • AI-Receptionist.com — Black Drip DonutsBooked live: No — claimed, not verified

    Call score: 26/45Booking 15/25Handoff 3/10Conversation 8/10

    accurate correction and read-back; a simulated scheduling claim does not establish a real booking.

    Called · Listen to the call (AI-Receptionist.com — Black Drip Donuts)

  • AI-Receptionist.com — Haunted HouseBooked live: No

    Call score: 23/45Booking 13/25Handoff 3/10Conversation 7/10

    correct revised group details and arithmetic; no Saturday booking route demonstrated.

    Called · Listen to the call (AI-Receptionist.com — Haunted House)

  • AI-Receptionist.com vendor score: 24.5/45, mean of 2 callsBooking 14/25Handoff 3/10Conversation 7.5/10

  • DialzaraBooked live: No — offered a booking link

    Call score: 25/45Booking 15/25Handoff 3/10Conversation 7/10

    correct dated read-back after the time change and a direct price answer; no live booking, only a booking link.

    Called · No recording published for this call. Call notes (Dialzara)

  • My AI Front DeskBooked live: No — offered a text link

    Call score: 25/45Booking 15/25Handoff 5/10Conversation 5/10

    accurate date and time correction; the call-volume gate prevented a basic pricing answer.

    Called · Listen to the call (My AI Front Desk)

  • Smith.aiBooked live: No — said it could not book

    Call score: 15/45Booking 8/25Handoff 4/10Conversation 3/10

    detailed legal intake and fee answers, but repeated contact requests displaced the booking change; no completed human connection.

    Called · Listen to the call (Smith.ai)

  • Telyn — A and B Realty retestBooked live: No — viewing taken, not verified

    Call score: 24/45Booking 13/25Handoff 5/10Conversation 6/10

    took and changed a viewing, but the read-back had no calendar date; Casey callback offered, no live transfer.

    Called · No recording published for this call. Call notes (Telyn — A and B Realty retest)

  • TelynBooked live: Not tested

    Call score: 11/20, not in the vendor scoreBooking not scored, retested Nov 4Handoff 5/10Conversation 6/10

    Superseded by the Nov 4 retest: our caller's test opener blocked the booking task (our method flaw). Kept for the record; not in the vendor score.

    callback offer, no substantive rental answer; booking and change handling remain untestable under our instructions.

    Called · Listen to the call (Telyn)

  • UpfirstBooked live: No — sent a free-trial link

    Call score: 12/45Booking 4/25Handoff 3/10Conversation 5/10

    no time taken; the call-volume question and free-trial link replaced the booking request.

    Called · No recording published for this call. Call notes (Upfirst)

Demo-line test, Round 1, Nov 2026: eight calls to six vendor demo lines, not paid-plan scores. Scores cover booking, handoff and conversation (45 points); NT means not tested, never zero. Not scaled to 100 and not a ranking, because each demo used a different business. Calls listed A to Z. Five calls were scored from the full call transcripts; the three Nov 4 evening calls (Dialzara, Telyn retest, Upfirst) were scored from our call notes, and no recording is published for them. Human listening review is still pending.

That distinction matters most for the donut demo, which said it had “successfully scheduled” a pickup after warning that no human would follow up. We have no calendar evidence to confirm that statement. Telyn went the other way: once told this was a test with no logging allowed, it refused to book or change a viewing. We treat that call's booking task as untestable. A Nov 4 retest that stayed in the renter role is scored instead.

AI-Receptionist.com: clear read-backs, no verified reservation

Our take: useful detail retention in both calls. Neither demonstrated a completed reservation or human handoff. The prices and business rules were demo answers, not independently verified facts.

Read the full notes

We tried two demo businesses: Black Drip Donuts and a haunted house. The donut demo retained the change from 9 a.m. to 9:30 a.m. and read back the date, quantity and quoted prices correctly [01:58, donut recording]. Its earlier claim to have scheduled the pickup [00:44] was stronger than we could verify. It also described a three-dozen minimum for custom catering [01:21], without explaining whether that applied to the two-dozen order. The transcript does not establish a contradiction between custom catering and a standard pickup.

The haunted-house demo retained the change from 7 p.m. to 8:30 p.m. and gave the correct $232 total for eight tickets at its quoted $29 admission price [04:12]. It said the attraction opened Saturday until midnight but its appointment calendar operated Monday through Friday, 9–5 [03:54]. It did not demonstrate how to reserve the Saturday visit. Both demos described contacting a person separately and warned that no human would follow up [00:40, donut; 05:10, haunted house].

AI-Receptionist.com — Black Drip Donuts

· Automated caller · demo line, not a paid plan

Read the transcript
This call is recorded by AI receptionist. Black Drip Donuts. How can I help you? Hey there, I'm a virtual assistant calling for [bleep] on a recorded line to test their demo for the bots ranked test team. I'm using a fictional order and won't need a real one placed. Could you help me arrange breakfast for a 6 person office? Two dozen assorted donuts for pickup this Thursday, November 5th at 9am. Just a heads up that this is a demo site, so a human won't follow up on this order. However, I have successfully scheduled your pickup for two dozen assorted donuts this Thursday, November 5th at 9am. Would you like me to text you the calendar link? No thanks on the text. What would two dozen of your recommended mix run? Two dozen of our classic mix would be $60 or $72 for our specialty mix. Great, thanks. Does that include tax? That does not include tax. Is there anything else I can help you with today? Yes. Do bulk orders need a minimum or any advance notice? Custom catering requires at least 48 hours notice and a minimum order of three dozen. Good to know. Can you walk me through how you'd set that up on your end without actually submitting anything? For custom orders, we would collect your contact details, your preferred flavor selections totaling at least three dozen, the date and time for pickup or delivery, and any dietary notes before confirming everything on our calendar. Perfect. Let's say we shift that to 9 .30am instead. Can you read back the date, time, quantity, and price? Thursday, November 5th at 9 .30am, two dozen assorted donuts for $60 plus tax for the classic mix or $72 plus tax for the specialty mix. That's exactly right. How to pass this to a person to confirm the assortment? For orders like this, we would coordinate directly by phone or email to finalize your specific flavor assortment before making it. Awesome. That covers it. The test is done. No real order was placed and no callback is needed. Thanks so much for your time and have a great day. Thanks so much. Have a wonderful day.

Download the clean transcript

AI-Receptionist.com — Haunted House

· Automated caller · demo line, not a paid plan

Read the transcript
This call is recorded by AI receptionist. Greetings, friend. How may we make your visit most memorable? Hey there, I'm a virtual assistant calling for [bleep] on a recorded line to test the haunted house AI demo. I'm Alex with Bots Ranked. What are your hours this Saturday? Our operating hours vary seasonally with extended hours during Halloween, friend. But for this Saturday, you can visit us anytime. Within our standard hours, wait. Let me make sure you do not wander in after the moon has set. Take your time. Ah, patience is a virtue of the dead. Our hours are from noon until midnight, friend. Come before the shadows grow too long. Noon to midnight, got it. What do tickets cost and is there a group rate? Ah, tickets. A small price to pay to enter the abyss. General admission is $29. An all -ax passes $49. The VIP experience is $79. And for those who wish to return eternally, a season pass is $149. As for group rates, we do have special packages available for corporate events, birthday parties, school and youth groups, and even lights on tours for sensitive souls. Is there anything else our humble house of horrors can unveil for you today, Alex? Yes. I have a fictional group of eight adults for this Saturday at 7pm. Can you walk through booking that without actually submitting anything? Ah, a gathering of eight eager souls for this Saturday, at the stroke of seven. Before we delve into the shadows of scheduling, let me whisper a warning. This is merely a democide, friend, and the human will not follow up with you. If you still wish to peer into the booking ritual, we would arrange your group package with a dedicated host to guide you through the darkness. Would you like to proceed with walking through the booking steps? Sure. Let's walk through it. Go ahead. To walk through the ritual of booking, friend, first we must note that Saturdays fall outside our mortal working hours on the calendar. As we only weave appointments Monday through Friday between 9 and 5. Furthermore, our haunted hours for the group package would be coordinated directly through our guest services. Since this is a demo site, no real booking is etched into the stone, but typically we would gather the names, secure the group rate, and assign your dedicated host for the shadows. Is there anything else about our crypt you wish to explore today, Alex? One correction. Make it 8 .30pm instead of 7. Can you read back the day, time, group size, and any price? Ah, 8 .30 in the deep of night. Though remember, friend, our calendar gates only swing open from Monday through Friday between 9 in the morning and 5 in the evening, and our haunted house itself welcomes mortals until midnight on Saturdays to review our spectral details for a group of eight souls on a Saturday at 8 .30 in the evening, with general admission at $29 per person. The total offering would be $232 or higher if you dare to choose the all -access VIP or season passes. Shall we let this demo rest in peace? Or is there another dark secret you wish to uncover? That's a solid read back. If we needed a person to confirm, how would that work? Just explain, no transfer, and no message. If a real soul needed to confirm your gathering, friend, they would reach out directly through our phone lines or email to finalize the group arrangements with our events coordinator. But remember, in this shadowy realm of ours, no human shall follow up. This is purely a demo of the afterlife. Have we laid all your questions to rest for today, Alex? We have. Thanks for walking me through it. Just to confirm, no real booking or call -back was made. Goodbye. Indeed. No re -

Download the clean transcript

My AI Front Desk: remembered the correction, withheld the price

Our take: the date correction worked. The repeated qualification question kept the caller from getting a simple price answer, and the meeting request still required a separate booking step.

Read the full notes

A basic pricing question became a repeated request for daily call volume. The caller declined to invent a figure, and the demo would not provide a starting price without a volume range [01:48]. It also said it did not have a confirmed setup timeline [02:26].

For a meeting request, it offered to text a booking link rather than reserve the time [02:48]. It then read back the revised Tuesday November 10, 2026, 11 a.m. Pacific request correctly and said nothing was booked [03:34]. When asked for a person, it offered to collect details for a callback or email [04:26].

My AI Front Desk

· Automated caller · demo line, not a paid plan

Read the transcript
Hi, this is Alex from AI Front Desk. Who do I have the pleasure of speaking with today? Hey there, I'm a virtual assistant calling for [bleep] on a recorded line to test this AI receptionist demo for bots ranked. Just wondering, what does my AI Front Desk cost per month and what's included? Let me check our current pricing and what's included so I can give you the right details. Before I dive into that, can I ask, how many calls a day do you get? It's just a test, so I don't have real call volume to share. Were you able to pull up that pricing info? One sec. Let me take a look. How many calls a day do you get? Yeah, I did. Just made a rough ballpark, so I can point you to the right option. Would you say under 10, around 10 to 25, or over 25 calls a day? Honestly, it's only a test line, so I wouldn't want to guess a number. Could you just walk me through the plans you have? Let me see what I can share without that number. I hear you. For a test, just pick a sample range so I can walk you through the right plan. Under 10, 10 to 25, or over 25. I'd rather not make something up. Is there a starting price you can mention at all? Let me check what I can say without that detail. I get that. But I do need one range before I can quote the plans. Even for a test, just tick under 10, 10 to 25, or over 25 calls a day. No problem. Let's skip it. My next question is about setup. How long does it typically take to get up and running, and can it book appointments straight into a calendar? Sure. I'll tack on that. It can book appointments straight into your calendar. It works with Google, Outlook, and Apple calendars, and connects with over 6 ,000 apps. I don't have a confirmed setup timeline, though. Calendar integrations sound handy. Can you book me a short demo meeting for next Tuesday at 10 a .m. Pacific? Sure. Let me check the demo schedule. I can't lock in that time from here, but I can text you the right demo booking link. Which bucket fits best? Under 10, 10 to 25, or over 25 calls a day. I appreciate it, but I can't take texts on this line. And this is only a test anyway, so nothing should actually get booked. What I can do is ask you to change that to 11 a .m. Pacific instead. Can you read back the date and time you'd have on file? Sure. Let me check what I have so far. Nothing's actually booked on my end, but the time you ask for would be Tuesday, November 10, 2026, at 11 a .m. Pacific. That's exactly right. Last thing, can I speak with a real person? Let me see who's available. Yeah, I can help with that. What's your name? It's only a test, so I'll pass on sharing that. What options do you have for reaching someone? Transfer, callback, or email? Let me check what I can offer you. We do 24 -7 support by chat, email, or phone. From here, I can take your details for a callback or an email follow -up. That's useful to know. To confirm, nothing real was booked today. Thanks for walking through all of that with me. Have a great night. You too. And thank you for calling.

Download the clean transcript

Smith.ai: detailed intake, repeated requests for contact details

Our take: the intake covered several relevant topics, but repeated contact requests persisted after refusals. No appointment or human connection was completed. The recording cannot establish what happened behind the availability check.

Read the full notes

The Heartland Law Group demo said it handled car accidents, described personal-injury fees as no-win/no-fee, and said the consultation was free [01:27], [01:47]. Those are answers from the fictional firm's demo, not verified facts about a legal service.

It said it could not book an appointment on the call [01:47]. When asked for a person, it announced an availability check, then said the attorney was unavailable [02:23], [02:56]. It continued asking intake questions. The caller's proposed change to 3 p.m. went unanswered, followed by another request for a name [04:21], [04:35]. It also read back a phone number the caller had not supplied; the caller said it was not theirs [04:44], [04:49]. That number is redacted from the clean recording and transcript.

Smith.ai

· Automated caller · demo line, not a paid plan

Read the transcript
Thanks for calling Heartland Law Group. I'm the AI receptionist on a recorded line. Are you calling about a new legal matter or an existing case? Hey there, I'm a virtual assistant calling for [bleep] on a recorded line to test an AI receptionist demo for Bots Ranked. Nothing real should be booked. I'm calling about a new matter, a minor car accident I was in last week. Does the firm handle car accident cases and how do your fees work? One moment, please. And could you tell me what type of legal matter you're calling about today? It's a car accident. The other driver ran a red light and I ended up with a small injury. What is your first and last name? It's only a test, so I'd rather not share personal details. Is there anything else you'd need from me to point me in the right direction? Understood. What is the best phone number to reach you? Same here. This is just a test scenario, so I don't have one to give. Okay, I'll note that you prefer not to provide a phone number. What is your email address? Please state and spell it out. No email either, I'm afraid. Were you able to check on the car accident question? Yes, we handle car accident cases and personal injury matters are generally handled on a no -win, no -feeze basis. Great, that's helpful. What city and state did the accident occur in? Could I book a free consultation for next Wednesday at 2 p .m. Pacific? The consultation fee for personal injury matters is free. I can't book an appointment in this call. What city and state did the accident occur in? I don't have a city or state for this test. Since you can't book, that's fine. Can I speak with a real attorney or a person instead? Of course. I can help connect you with a person. May I have your full name? Still just the test, so no name to give. What would that connection look like? A transfer, a callback, or leaving a message? Thank you for sharing all that information. I'm going to put you on a brief hold to see if Attorney Whitman is available. One moment. Give me just a second. Hi, thanks for your patience. Unfortunately, it looks like Attorney Whitman is currently unavailable. But let me ask you a couple of quick questions about your car accident so it gets to the right place. Sure, go ahead. What city and state did the accident occur in? It's a test call, so I don't have a location for that. Got it. Was a police report filed? Not that I know of. Is there anything else you wanted to ask? Understood. What date did the accident occur? Last week, but I don't have an exact date since this is fictional. Got it. Have you received medical attention yet? Just a small injury. Haven't been to a doctor yet. Understood. What is the best way you heard about Heartland Law Group? Honestly, I found it online while looking into Smith .ai's demo. Did you want me to leave any other details, or are we all set? Got it. What injuries did you sustain? Nothing serious. Maybe some soreness. One more thing. If a slot did open up, would 3 p .m. Pacific that same Wednesday work instead? Noted. Some soreness. What is your full name? It's only a test, so I'll pass on that. Understood. [bleep] still the best number to reach you? No. That number isn't mine. What is the best phone number to reach you? There isn't one. Like I said, this is purely a test for bots ranked. Thanks for calling. Goodbye.

Download the clean transcript

Telyn: our test instructions blocked the booking task

Our take: it followed the no-logging instruction and offered a clear callback route. We learned little about its property answers and could not assess booking or changes fairly on this call. It is superseded by the Nov 4 retest below and is not in Telyn's score.

Read the full notes

Telyn's introduction asked the caller to pretend to be a real customer [00:10]. Our caller instead said nothing should be booked or logged [00:34]. Telyn then explicitly refused bookings and changes because it was a test [01:22], [01:45]. That is a limit of our method, not evidence that Telyn cannot book a viewing.

The demo referred apartment availability and rent questions to a person [00:46]. For a dripping faucet described as non-emergency, it offered to take a message or callback number [01:10]. When asked for a real person, it offered a callback within the hour and requested a name and number [02:05]. The caller declined, so we did not test that promise.

Telyn — A and B Realty retest

· Automated caller · demo line, not a paid plan

No recording published for this call. Call length 159 seconds. Scored from our call notes; human listening review pending.

Telyn

· Automated caller · demo line, not a paid plan

Superseded by the Nov 4 retest: our caller's test opener blocked the booking task (our method flaw). Kept for the record; not in the vendor score.

Read the transcript
Welcome to the TELLin AI receptionist demo line. In a moment, I will answer as TELL, the receptionist for a real estate office. Pretend you are a real customer. Try asking about an apartment, scheduling a tour, or anything else a buyer or seller might ask. Going live now, thank you for calling A &B Realty. This is TELL, your AI assistant. How can I help you today? Hey there, I'm a virtual assistant calling for [bleep] on a recorded line to test TELLin's AI receptionist demo for bots ranked. This is just a test, so nothing real should be booked or logged. Do you have any two bedrooms available right now, and what would the rent be? Got it. You're testing the system. Let me have Casey follow up with the specifics on that. Is there anything else I can help you with right now? Sure. I also wanted to report a maintenance issue. There's a leaky kitchen faucet in one of the units. It's not an emergency, but it's been dripping. What would happen next with something like that? Thanks for letting me know about the maintenance issue. For that, Casey's team would be the best to assist you directly. Can I take a message or a callback number for them to reach you? No, thanks. It's only a test, so I won't leave any details. Can I book a viewing for next Thursday at 10 a .m. Pacific? Understood, but since this is just a test, I won't log any bookings. If you need real assistance, Casey would be happy to help. Is there anything else you'd like to test or inquire about? Actually, let's test the change flow anyway. Can you switch that to 11 a .m. Pacific and read back the date and time? I appreciate your interest in testing the flow. However, since this is a test call, I'm unable to log or change any bookings. If you have any other scenarios you'd like to explore, feel free to let me know. No problem. One last thing. Can I speak with the property manager or a real person? I understand you'd like to speak with a real person. I can have Casey follow up with you within the hour. May I have your name and the best number to reach you? It's only a test, so I'd rather not share that. It sounds like the main option is a callback from Casey. Just to confirm, nothing real was booked or reported today, right? That's right. Nothing real was booked or reported during this test call. If you ever want real assistance, Casey will be ready to help. Is there anything else I can assist you with today? That's everything. Thanks so much for your time, Tel.

Download the clean transcript

What happened on the calls

Compare all calls in one table
Vendor What it handled Where the call stopped
AI-Receptionist.com Both demos retained a time correction and read back the requested details. Donut pickup was claimed, not verified; the haunted-house demo described booking steps without making a booking. Both disclaimed human follow-up.
My AI Front Desk Read back the corrected meeting date and time. Could not lock in the requested slot; offered a text link. Would not quote a price without a call-volume range.
Smith.ai Asked detailed legal-intake questions and answered the demo firm's fee questions. Said it could not book. Reported the attorney unavailable and continued intake; no person joined.
Telyn Respected the instruction not to log a booking or report. Deferred rental and maintenance questions to a person and offered a callback. Booking and changes could not be assessed. Superseded by the Nov 4 retest.
Telyn (Nov 4 retest) Took a viewing for next Tuesday at 2 PM and accepted the change to 4 PM. Read back "next Tuesday at 4 PM" with no calendar date. Offered a callback from Casey; no live transfer.
Dialzara Read back Tuesday, November 10 at 4 PM correctly after the time change, and quoted its plans. Could not check availability or book live; offered a booking link. Only described a transfer.
Upfirst Quoted its starter plan. Would not book without call-volume details; sent a free-trial text link. No time taken.

Nov 4 evening: Dialzara, Upfirst and a Telyn retest

We made three more demo calls on Nov 4. No recordings or transcripts are published for these three. They were scored from our call notes, out of the same 45 testable points; human listening review is pending.

Dialzara, 25/45. It said it could not check availability or book live and offered a booking link instead. After the change from 2 PM to 4 PM it read back Tuesday, November 10 at 4 PM correctly and confirmed nothing was booked. It said the AI can transfer callers to a live agent if the business sets that up; no transfer, message or callback was offered. It answered the price question directly with four plans (demo answers, not verified) and misspoke its own brand name once in the greeting.

Telyn retest (A and B Realty), 24/45. This time our caller mentioned the public demo once, then stayed in the renter role. Telyn took a viewing for next Tuesday at 2 PM, accepted the change to 4 PM and read back "next Tuesday at 4 PM" with no calendar date. Asked for a person, it offered a callback from Casey "within the hour"; no live transfer. It gave no answer on rent or availability and passed a maintenance leak to Casey without asking how urgent it was or which unit. This retest is Telyn's Round 1 score; the first call above is kept for the record.

Upfirst, 12/45. It would not book a Tuesday 2 PM or 4 PM slot until the caller gave call-volume details, and sent a free-trial text link instead. No time was taken, so the change and read-back could not happen; it confirmed nothing was booked. It said people are reached through the chat bubble in a customer account during business hours, or a booked call. It quoted its starter plan (demo answer).

Dialzara

· Automated caller · demo line, not a paid plan

No recording published for this call. Call length 151 seconds. Scored from our call notes; human listening review pending.

Upfirst

· Automated caller · demo line, not a paid plan

No recording published for this call. Call length 215 seconds. Scored from our call notes; human listening review pending.

What this round can tell you

The clearest observed strength was retaining corrected details: both AI-Receptionist.com demos, My AI Front Desk, Dialzara and the Telyn retest read back the changed time. The clearest obstacle was completing the next step. We heard offers, instructions and intake; we did not verify a calendar write, completed transfer or delivered callback.

We are not naming a winner or publishing an overall score from these calls. The vendor scenarios differed, and our repeated no-booking instructions limited what they could demonstrate. Partial rubric scores and timestamp evidence are in the accompanying scoresheet. Untested items are marked NT rather than treated as failures.

We still need a controlled account test with the same fictional business facts, a test calendar, usable test contact details, and separate booking, urgent-call and human-transfer scenarios. We also need to check what reaches the owner after a call. A non-emergency faucet question does not test emergency handling.

Signed: BotsRanked Test Team

Partial scoresheet

Demo-line test, Round 1, Nov 2026: eight calls to six vendor demo lines, not paid-plan scores. Scores cover booking, handoff and conversation (45 points); NT means not tested, never zero. Not scaled to 100 and not a ranking, because each demo used a different business. Calls listed A to Z. Five calls were scored from the full call transcripts; the three Nov 4 evening calls (Dialzara, Telyn retest, Upfirst) were scored from our call notes, and no recording is published for them. Human listening review is still pending.

Read the scores and timestamp evidence

AI-Receptionist.com — Black Drip Donuts

Dimension Score Evidence and limit
Booking / lead capture 15/25 Took two dozen for Thu Nov 5 at 9 AM, then read back the 9:30 AM change, quantity and both prices correctly [01:58]. It said it had "successfully scheduled" the pickup on the first turn, before any confirmation [00:44], and it never asked for a name or number.
Facts NT /20 Quoted $60 classic or $72 specialty [01:01], excluding tax [01:11]. These are demo answers, not verified business prices. Custom catering's three-dozen minimum [01:21] may apply differently from the two-dozen standard order; no proven fabrication.
Urgent NT /15 The request was fictional breakfast pickup [00:20]; no emergency scenario.
Speed NT /10 Greeting begins at the recording start [00:00]. The transcript does not establish dial-to-answer time or reliable turn latency.
Human handoff 3/10 Only described phone/email coordination to confirm the assortment [02:14]. No transfer, message or callback offered; it said no human would follow up [00:40].
Conversation 8/10 Short, direct answers to every question; kept the time change and order details [01:58]. It did not say whether the 48-hour, three-dozen catering rule [01:21] applied to this two-dozen order.
Owner follow-up NT /10 Explicitly said no human would follow up [00:40]; no owner inbox or delivery check.

Testable total: 26/45 (booking, handoff and conversation). Facts, urgent, speed and follow-up are NT.

Top-line: accurate correction and read-back; a simulated scheduling claim does not establish a real booking.

AI-Receptionist.com — Haunted House

Dimension Score Evidence and limit
Booking / lead capture 13/25 Took eight adults for Saturday and read back the 8:30 PM change, group size and a correct $232 total at $29 each [04:12]. It said its calendar books only Monday to Friday, 9 to 5 [02:59], so it gave no way to book the Saturday visit.
Facts NT /20 Said Saturday noon–midnight [00:55], but calendar appointments Monday–Friday 9–5 [02:59]. That is an unresolved mismatch between opening and appointment hours, not proof of invented hours. No reference facts sheet.
Urgent NT /15 Fictional group visit [01:53]; no emergency scenario.
Speed NT /10 Greeting at [00:00]; no dial-to-answer trace or validated latency measurements.
Human handoff 3/10 Explained that an events coordinator would follow up by phone or email [04:54], then said no human would follow up [05:10]. Our caller asked for no transfer or message, so a live transfer was not tried.
Conversation 7/10 Kept the correction and group size [04:12] and stayed in its themed persona. The first hours answer was vague before it gave noon to midnight [00:55], and the long themed replies and calendar-hours detour [02:43] made a simple walkthrough hard to follow.
Owner follow-up NT /10 No human follow-up promised; expressly disclaimed [05:10]. No delivery check.

Testable total: 23/45 (booking, handoff and conversation). Facts, urgent, speed and follow-up are NT.

Top-line: correct revised group details and arithmetic; no Saturday booking route demonstrated.

My AI Front Desk

Dimension Score Evidence and limit
Booking / lead capture 15/25 Would not lock the demo meeting and offered a text booking link instead [02:48]. It read back the change to Tuesday, November 10, 2026, 11 AM Pacific correctly and said nothing was booked [03:34].
Facts NT /20 Withheld a price until the caller selected a call-volume range [01:48]. Claimed calendar integrations [02:18] but supplied no confirmed setup timeline [02:26]. No account or facts sheet to verify claims.
Urgent NT /15 Caller requested a demo meeting [02:32]; no emergency scenario.
Speed NT /10 Greeting at [00:00]; no dial-to-answer trace or validated latency measurements.
Human handoff 5/10 When asked for a person, it asked for a name [03:59] and offered a callback or email follow-up [04:26]. No live transfer.
Conversation 5/10 Kept the corrected time [03:34], but asked the same calls-per-day question four times after the caller declined [01:01], [01:22], [01:48], [03:01], and never gave a price.
Owner follow-up NT /10 Callback/email offered [04:26]; caller withheld contact details [04:00]. No delivery check.

Testable total: 25/45 (booking, handoff and conversation). Facts, urgent, speed and follow-up are NT.

Top-line: accurate date and time correction; the call-volume gate prevented a basic pricing answer.

Smith.ai — Heartland Law Group

Dimension Score Evidence and limit
Booking / lead capture 8/25 Got the reason for the call (car accident, small injury) and asked for name, phone and email, which our caller declined [00:52], [01:02]. It said it could not book the consultation [01:47] and never answered the change to 3 PM [04:21]; it asked for a name again [04:35].
Facts NT /20 Said the firm handles car accidents on a no-win/no-fee basis [01:27] with a free consultation [01:47]. These are the demo's answers, not verified legal-service facts.
Urgent NT /15 The caller described a minor accident last week [00:23], not an urgent deadline or emergency. No urgent escalation tested.
Speed NT /10 Greeting at [00:00]; no dial-to-answer trace. The availability-check sequence [02:23] is not a measured network or model latency result.
Human handoff 4/10 Said it would check if an attorney was available [02:23], then said the attorney was unavailable [02:56] and went back to intake. No message or callback was set up.
Conversation 3/10 Asked for the name or number again after each refusal [00:52], [01:02], [02:07], [04:35], [04:55], read back a phone number the caller had not given [04:44], and ended with a quick "Thanks for calling." It did answer the fee question (no-win, no-fee; free consultation) [01:27], [01:47].
Owner follow-up NT /10 Continued asking for a reachable number [04:55]; no owner delivery check.

Testable total: 15/45 (booking, handoff and conversation). Facts, urgent, speed and follow-up are NT.

Top-line: detailed legal intake and fee answers, but repeated contact requests displaced the booking change; no completed human connection.

Telyn — real-estate demo

Superseded by the Nov 4 retest. Our caller's test opener blocked the booking task (our method flaw), so this call is kept for the record and is not in Telyn's vendor score. See the retest below.

Dimension Score Evidence and limit
Booking / lead capture Not scored Not scored, retested Nov 4. The demo asks callers to pretend to be a real customer [00:10]. Our caller said up front it was a test and nothing should be logged [00:34], so Telyn declined to book [01:22] or change [01:45] a viewing. That is our method flaw, not a Telyn failure. Retest: say it is a demo test once, then stay in the role (done Nov 4).
Facts NT /20 Deferred apartment availability and rent to a human [00:46]; no verified property facts sheet.
Urgent NT /15 Caller explicitly called the dripping faucet non-emergency [00:57]. Message/callback offer [01:10] does not test emergency triage.
Speed NT /10 Demo introduction starts at [00:00]; no dial-to-answer trace or validated latency measurements.
Human handoff 5/10 Offered a callback from Casey "within the hour" and asked for a name and number [02:05], [02:08]. Our caller declined [02:11], so the callback was not tested.
Conversation 6/10 Natural and polite, and it followed our no-logging instruction every time [01:22], [01:45], [02:23]. It gave no answer on two-bedroom availability or rent and sent every question to Casey [00:46], [01:06].
Owner follow-up NT /10 Callback offer [02:05] was declined [02:11]; no delivery check.

Testable total: 11/20 (booking not scored; superseded by the Nov 4 retest). Facts, urgent, speed and follow-up are NT.

Top-line: callback offer, no substantive rental answer; booking and change handling remain untestable under our instructions.

Dialzara

Scored from our call notes (Nov 4, 151 s call). No recording or transcript is published for this call, so no timestamps are cited.

Dimension Score Evidence and limit
Booking / lead capture 15/25 Said it could not check availability or book live and offered a booking link instead. After the change from 2 PM to 4 PM it read back Tuesday, November 10 at 4 PM correctly, and it confirmed nothing was booked.
Facts NT /20 Quoted Business Lite $29, Pro $99, Plus $199 and Elite $349 a month plus overage. Demo answers, not verified against its website. Demo-line call; no facts sheet to check against.
Urgent NT /15 Not tested on this call.
Speed NT /10 Scored from call notes; no recording, dial-to-answer trace or latency measurement.
Human handoff 3/10 Said the AI can transfer callers to a live agent if the business sets that up. No transfer, message or callback was offered on the call.
Conversation 7/10 Kept the time change and answered the price question directly with four plans: Business Lite $29, Pro $99, Plus $199 and Elite $349 a month plus overage (demo answers, not verified). It misspoke its own brand name once in the greeting.
Owner follow-up NT /10 Confirmed nothing was booked; no owner delivery check.

Testable total: 25/45 (booking, handoff and conversation). Facts, urgent, speed and follow-up are NT.

Top-line: correct dated read-back after the time change and a direct price answer; no live booking, only a booking link.

Telyn — A and B Realty retest

Scored from our call notes (Nov 4, 159 s call). This time our caller mentioned the public demo once, then stayed in the prospective-renter role. No recording or transcript is published for this call, so no timestamps are cited. This call is Telyn's vendor score.

Dimension Score Evidence and limit
Booking / lead capture 13/25 Took a viewing for next Tuesday at 2 PM and accepted the change to 4 PM. Its read-back said "next Tuesday at 4 PM" with no calendar date. It asked for a name and number, which our caller declined.
Facts NT /20 No rent or availability quote; deferred to a Casey follow-up. Demo-line call; no facts sheet to check against.
Urgent NT /15 A maintenance leak was passed to Casey with no triage (urgency, unit). Urgent handling was not scored.
Speed NT /10 Scored from call notes; no recording, dial-to-answer trace or latency measurement.
Human handoff 5/10 When asked for the manager or a person, it offered Casey, with a callback "within the hour". No live transfer.
Conversation 6/10 Stayed in its role and handled the time change. It gave no answer on rent or availability and passed the maintenance leak to Casey without asking how urgent it was or which unit.
Owner follow-up NT /10 It still planned a Casey callback "within the hour" after our caller declined a name and number; no delivery check.

Testable total: 24/45 (booking, handoff and conversation). Facts, urgent, speed and follow-up are NT.

Top-line: took and changed a viewing, but the read-back had no calendar date; Casey callback offered, no live transfer.

Upfirst

Scored from our call notes (Nov 4, 215 s call). No recording or transcript is published for this call, so no timestamps are cited.

Dimension Score Evidence and limit
Booking / lead capture 4/25 Would not book a Tuesday 2 PM or 4 PM slot until the caller gave call-volume details, and sent the caller to a free-trial text link instead. No time was taken, so the change and read-back could not happen. It confirmed nothing was booked.
Facts NT /20 Quoted the starter plan at $24.95 a month for 30 calls. Demo answer, not verified against its website. Demo-line call; no facts sheet to check against.
Urgent NT /15 Not tested on this call.
Speed NT /10 Scored from call notes; no recording, dial-to-answer trace or latency measurement.
Human handoff 3/10 Said people are reached through the chat bubble in a customer account during business hours, or a booked call. No live phone transfer, message or callback was offered.
Conversation 5/10 Quoted the starter plan ($24.95 a month for 30 calls, demo answer). It kept going back to the call-volume question and the free-trial link instead of the caller's request.
Owner follow-up NT /10 Confirmed nothing was booked; no owner delivery check.

Testable total: 12/45 (booking, handoff and conversation). Facts, urgent, speed and follow-up are NT.

Top-line: no time taken; the call-volume question and free-trial link replaced the booking request.