Buyer's guide, updated July 2026

The AI receptionist, priced honestly

An AI receptionist answers every call on the first ring, at roughly a fortieth of the published per-minute price of a human answering service. Ruby and PATLive publish plans that work out at $1.95 to $5.00 per included minute. DILR Voice is $0.14 a minute and starts free. That is real arithmetic on real rate cards, and below we show the working, the caveats, and the calls you should still send to a human.

Published rate cards, not estimatesThe stats we refused to useWhen to hire a human instead

The short answer

An AI receptionist is a voice agent that answers your inbound calls 24/7, works out why the person is calling, answers from your own documents, books the appointment, captures the details and hands the call to a human when it needs one. DILR Voice runs it as a multi-agent pipeline, so calls hand off between specialised agents without the caller repeating themselves.

On price it is not a close contest, and we will show the division. PATLive Pro is $1,170 a month for 600 minutes, which is $1.95 a minute. Ruby Starter is $250 for 50 minutes, which is $5.00. DILR Voice is $0.14. Both of those human numbers come off their own live pricing pages.

Price is not the whole decision, though. A human answering service is doing a different job, and there are calls that should never meet a bot. We have put those in their own section rather than burying them, because a page that pretends otherwise is not worth your time.

The maths, shown

What a minute costs, on published rate cards

Every human figure below is that provider's own monthly plan price divided by the minutes that plan includes. No modelling, no estimates, no assumptions. Just division on numbers anyone can go and read.

Effective cost per included minute

Monthly plan price divided by included minutes, from each provider's live pricing page in July 2026. DILR Voice is a pure usage rate, so it has no allowance to divide.

Provider and planMonthlyIncluded minutesPer included minutePublished overage
DILR Voice, GrowthNo plan feePay per minute$0.14Not applicable
Ruby Starter$25050$5.00Not published
Ruby Small Business$395100$3.95Not published
Ruby Popular$720200$3.60Not published
Ruby Growing Business$1,725500$3.45Not published
PATLive Starter$25075$3.33$2.35/min
PATLive Standard$460200$2.30$2.20/min
PATLive Premium$720350$2.06$2.10/min
PATLive Pro$1,170600$1.95$2.00/min

Sources: ruby.com pricing and patlive.com pricing, read in July 2026. Two caveats we are not going to bury. First, the effective rate assumes you use your whole allowance: if you use half your minutes, the real human rate doubles. Second, PATLive publishes overage at $2.00 to $2.35 a minute, which is the independent cross-check that this ballpark is right, and also a warning that going over is costly. UK readers: Face for Business publishes plans from £49 a month (ffb.co.uk pricing), but does not publish included minutes, so we cannot do this division for them honestly and have left them out.

Why answering matters

The 21x number, and the citation everyone gets wrong

The odds of qualifying a lead fall about 21 times when your response slips from 5 minutes to 30 minutes. There is a fourfold decline between 5 and 10 minutes alone, and a sixfold decline across the first hour. The study behind it covers three years of data across six companies, more than 15,000 leads and over 100,000 call attempts.

Now the part almost every page in this industry gets wrong. That figure is credited to Harvard Business Review nearly everywhere you will see it quoted. It is not HBR. It is the Lead Response Management study (Oldroyd, MIT Sloan with InsideSales.com), led by Dr James Oldroyd, then at MIT Sloan. HBR published a separate piece in 2011 with different numbers. We are pointing this out because if a vendor cannot get the citation right on the one statistic their entire pitch rests on, you should wonder what else they have not checked.

An AI receptionist does not try to win the 5-minute race. It answers on the first ring, which takes the response-time curve off the table completely.

On missed calls specifically, the honest picture is thinner than the internet suggests. A 30-day study of 85 businesses across 58 industries found only 37.8% of calls were answered by a live person, with 24.3% getting no response at all (411 Locals, January 2016). Note the date: that is a 2016 study, and it is routinely passed off as current. Separately, Invoca platform datareports 27% of calls to home-services businesses go unanswered, which is their own platform data rather than independent research. We have left out the famous "$126,000 a year lost to missed calls" and "85% never call back" figures entirely: neither traces to a real source.

Hear it answer your actual calls

Book 30 minutes and we will put an agent on a real call flow from your business, then tell you honestly whether the calls you get should be going to a human instead.

The honest limits

When you should hire a human answering service instead

We sell an AI receptionist. Here is when not to buy one. This section is not a formality, it is the part of the page we would want to read.

Buy a human service

Ruby, PATLive, Moneypenny, Face for Business and the rest

When the call is the product

If your callers are distressed, if the conversation needs judgment or negotiation, if a wrong answer is a legal or safety problem, or if a person picking up is genuinely your brand, buy a human service. The price difference is not the point and we are not going to argue you out of it.

When your volume is small

At ten calls a month the difference between $0.14 and $3.00 a minute is a rounding error on your P&L. Buy the human. AI receptionists earn their keep on volume, on out-of-hours cover and on repetitive calls, not on small numbers.

What DILR Voice does not do

The gaps we are not going to hide

No HIPAA, no SOC 2

DILR Voice does not claim HIPAA and does not publish a SOC 2 report. Retell AI and Vapi both offer HIPAA and we have said so on those pages. Our HIPAA-aligned work is Dilr Mira, private clinical models that run on your own hardware.

It is not a person

An AI receptionist is very good at the call that has a right answer, and it should hand over the moment a call does not. The design that works is not AI instead of your team, it is AI answering everything on the first ring and escalating what deserves a person.

By industry

The calls look different in your business

A valuation request and a patient booking are not the same call, and the research on each is different. We have written these up separately rather than swapping the keyword.

What people actually ask

Straight answers, including the Reddit questions

How much does an AI receptionist cost in 2026?
DILR Voice is $0.14 a minute on Growth, plus a monthly fee per phone number, and it starts free with $10 of trial credits and no credit card. For comparison, human answering services publish plans that work out between roughly $1.95 and $5.00 per included minute: PATLive Pro is $1,170 a month for 600 minutes, which is $1.95 a minute, and Ruby Starter is $250 for 50 minutes, which is $5.00 a minute. So on published rate cards an AI receptionist runs roughly 14 to 36 times cheaper per minute. That is the honest arithmetic, but it is not the whole story: a human service is doing a different job, and the human rate only holds if you use your full allowance.
Is an AI receptionist cheaper than a human answering service?
Per minute, yes, and it is not close. On published rate cards the cheapest human plan we found is PATLive Pro at $1,170 a month for 600 included minutes, or $1.95 a minute. The dearest is Ruby Starter at $250 for 50 minutes, or $5.00 a minute. DILR Voice is $0.14 a minute. Two honest caveats. First, that effective rate assumes you use your full monthly allowance; use half and the human rate doubles. Second, PATLive publishes overage at $2.00 to $2.35 a minute, so going over is expensive on either side of the comparison. The real question is not price, it is whether the calls you get need a human.
What does an AI receptionist actually do?
It answers every inbound call 24 hours a day, identifies why the person is calling, answers questions using your own documents through a RAG knowledge base, books appointments, captures the caller's details, and transfers to a human when the call needs one. After the call it can trigger follow-up automatically: a CRM record, a text, a callback task. DILR Voice runs this as a multi-agent pipeline, so a call can hand off between specialised agents while keeping context, rather than one bot trying to do everything.
What do people on Reddit say about AI receptionists?
We are not going to tell you, because we cannot verify it. Reddit is not machine-readable for us, so any quote or thread summary we published here would be invented, and inventing Reddit sentiment on a page about honest pricing would be self-defeating. We have linked live Reddit searches below so you can read the actual threads and judge for yourself. That is the honest version of a "what Reddit says" section, and you should be suspicious of any vendor page that offers you tidy quotes instead.
How fast do you have to answer a lead?
The best research on this found that the odds of qualifying a lead drop about 21 times when your response slides from 5 minutes to 30 minutes, with a fourfold decline between 5 and 10 minutes. It is based on three years of data across six companies, more than 15,000 leads and over 100,000 call attempts. Worth knowing: this study is credited to Harvard Business Review almost everywhere online, and that is wrong. It is the Lead Response Management study led by Dr James Oldroyd, then at MIT Sloan, with InsideSales.com. HBR published a separate 2011 piece with different numbers. An AI receptionist answers on the first ring, which is the only response time that beats the curve entirely.
When should I hire a human answering service instead?
When the call is the product. If your callers are distressed, if the conversation needs real judgment or negotiation, if a wrong answer is a legal or safety problem, or if your brand is built on a person picking up, buy a human service and do not think about it again. The same applies if you take very few calls: at ten calls a month the price difference is rounding, and a human is better. AI receptionists earn their place on volume, on out-of-hours cover, and on the repetitive calls that are eating your team, not on replacing judgment.
Can an AI receptionist transfer to a human?
Yes, and this is the setting that makes the whole thing work. DILR Voice does warm human transfer: the agent recognises when the call needs a person, hands it to the right one and passes the context so the caller does not start over. The right design for most businesses is not AI instead of humans, it is AI answering everything on the first ring and escalating the calls that deserve a person, which is exactly the set of calls your team should have been getting all along.
Is DILR Voice HIPAA compliant for medical calls?
No. DILR Voice does not claim HIPAA and we will not sign a BAA for it. If you need HIPAA on the voice platform itself, Retell AI and Vapi both offer it and we have said so on our comparison pages. What DILR does have for healthcare is Dilr Mira, a class of private clinical models that run on your own hardware so patient data never leaves your environment. For UK clinics and GP practices, where the questions are usually GDPR and NHS data handling rather than HIPAA, see our AI receptionist for clinics page.

We do not reproduce Reddit quotes, because we cannot link you to a thread we have verified. Read the live discussion and judge it yourself:

Put an AI receptionist on your real calls

Book 30 minutes and we will run a live inbound demo against your actual call flow. Or start free with $10 of credits and no credit card, and hear it yourself first.