Is Ruby Receptionist Legit? Worth It Next to an AI Receptionist

Coding Liquids tutorial cover featuring Sagnik Bhattacharya for Is Ruby Receptionist Legit? Worth It Next to an AI Receptionist.
Coding Liquids tutorial cover featuring Sagnik Bhattacharya for Is Ruby Receptionist Legit? Worth It Next to an AI Receptionist.

Ruby is a real, openly priced service, and it is worth paying for when callers need a person, calls run long or carry high-value enquiries. An AI receptionist is usually cheaper for simple, high-volume calls. Most small firms end up choosing by call mix, not by headline price.

"Legit" and "worth it" are different questions. Ruby's published prices show a real business with transparent plans. Whether it earns its cost depends on how many calls you get, how long they run, and how many need judgement rather than a booking. A firm taking 40 short booking calls a month and a firm taking 40 long, anxious enquiries can reach opposite answers on the same price list.

Follow me on Instagram@sagnikteaches

What Ruby publishes about its plans

On its pricing page, read on 5 November 2026, Ruby lists monthly virtual receptionist plans by minutes.

Connect on LinkedInSagnik Bhattacharya
PlanMonthly pricePrice per included minute
50 minutes$250$5.00
100 minutes$395$3.95
200 minutes$720$3.60
500 minutes$1,725$3.45

The page also lists live chat plans, and bundles that take 20 percent off the chat price when chat is added to a receptionist plan. All receptionist plans are described as including 24/7 live answering, scheduling and outbound call assistance, payment collection, lead qualification, call forwarding, voicemail transcription, an online portal and local or toll-free numbers. The page says optional AI enhancements are available at no extra cost and that there are no activation, onboarding or setup fees.

Subscribe on YouTube@sagnikteaches

What the page does not say matters just as much. It does not define what counts as a minute, does not state the overage rate, does not mention contract terms and does not mention a free trial. Ask for all four in writing before you sign. In particular, ask whether a call is billed from the moment the receptionist picks up, and whether time spent leaving a message or making an outbound call counts. Those answers change your cost per call more than the plan price does. The tutorial on Ruby's minute plans goes into those details.

The Complete Excel Course with AI Integration thumbnail

The Complete Excel Course with AI Integration

From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.

By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner

Buy now

Cost per call: the sum for your own phone

To compare a minute plan with a per-call service, convert your calls into minutes. You need three numbers: calls a month, average minutes per call, and the plan that fits.

Calls per month                         = C
Average receptionist minutes per call   = M
Minutes needed                          = C x M
Plan chosen                             = smallest plan with enough minutes
Cost per call                           = plan price / C

Say an electrical contractor with ten staff receives 40 calls a month that the receptionist would answer, averaging 2.5 minutes. That is 100 minutes, which fits the $395 plan exactly, so $395 divided by 40 is about $9.88 a call. If the calls average three minutes, the contractor needs 120 minutes, which is above the 100-minute plan and means either buying the 200-minute plan at $720 ($18 a call) or paying whatever overage Ruby charges. That cliff between plan sizes is the thing to model. A small change in call length can double your cost per call.

For a quick estimate from your own week of calls, use the estimating tutorial, which walks through counting minutes from one week of real calls.

What AI receptionists cost for the same 40 calls

Pricing for AI receptionists is usually per call or per minute, with a low monthly floor. Smith.ai is a useful benchmark because it sells both. Its main pricing page, read on 5 November 2026, lists call answering by real people at $300 for 30 calls, $810 for 90 and $2,100 for 300, with overage of $11.50, $10.50 and $8.50 a call respectively and a 10 percent discount for a 12-month subscription. Its separate AI receptionist page lists a free plan with 25 calls a month and then $3 a call, a Pro plan from $150 a month for 75 calls ($2 a call, then $2.50 for each extra call) and larger tiers down to $1.67 a call at 300 calls. The AI plans are month to month with 30 days' notice to cancel.

On the free AI plan, the contractor's 40 calls cost $45: 25 free, then 15 at $3. That is about $1.13 a call against Ruby's $9.88. At higher volume the free plan stops being the cheap option: 300 calls on it would cost $825, while the 300-call Pro tier at $1.67 a call comes to about $500. A human plan at that volume would be $2,100 for 300 calls on Smith.ai, or $7 a call.

The numbers are illustrative and the AI price may differ on your plan, but the shape is reliable. AI is much cheaper for simple calls and the gap widens with volume. Human reception costs more per call and is worth it where the call itself has value.

Which calls need a human

The better question is not "human or AI?" but "which calls are these?" Sort a week of calls into four groups.

Call typeExampleAI fitHuman fit
Simple and routineOpening hours, booking a site visit, reschedulingStrongFine but costly
Information-gatheringA new enquiry: name, address, what is wrong, how urgentGood if the script is clearStrong, asks follow-ups
High-value or sensitiveA prospective client with a complex matterWeakStrong
Upset or urgentA caller with a fault, a complaint or an emergencyPoorStrong

Here is an illustrative split for a four-fee-earner law firm over a month: 120 calls, of which 50 are routine (diary changes, document chasers), 35 are new enquiries, 25 are sensitive matters and 10 are distressed callers. The 50 routine calls suit AI. The 35 enquiries could go to either, depending on the script. The 35 sensitive and distressed calls need a person, and those are also the calls where a bad first impression costs the most.

Caller preference and what you can test

Preferences vary, and I would not assume either way. Some callers are happy to give an address and a time to an automated voice and appreciate the speed. Others hang up when they realise it is not a person, particularly if the first thing they hear is a clumsy greeting. The calls where callers least want automation are the ones where they are worried, confused or paying a lot.

You can test preference instead of guessing. Run the AI on routine calls for four weeks and track three things: calls abandoned before the AI finished, callers who asked for a person, and complaints. If more than one in ten callers asks for a human, the script is failing or the call type is wrong for AI. The tutorial on call scripts and escalation rules covers how to write the handover so a request for a person works first time.

A script that tells callers what is happening

"Thanks for calling [business name]. I'm an automated assistant.
I can book a visit, take a message or pass you to a person.
If you'd rather speak to someone now, say 'person' at any time."

The line "say 'person' at any time" matters more than anything else in the script. It tells the caller they are not trapped, and in practice it lowers the number who ask for it.

Hybrid options: the middle path

Many firms do best by mixing. Three patterns are common.

  • AI first, human on transfer. The AI answers and handles routine calls; anything it cannot handle transfers to your own staff or to a human service.
  • Human by day, AI by night. Your office answers during the day and an AI takes messages overnight and at weekends, which is cheaper than paying a human service for quiet hours.
  • Human service with AI extras. Ruby's page lists AI-assisted call flows and optional AI enhancements at no extra cost. Ask exactly what the AI does, whether it answers or only assists, and what is billed.

The right pattern follows your call mix. If most calls are routine, AI first. If most calls are valuable and few, a human service. If volume is spiky, a hybrid avoids paying human rates for quiet hours. For a closely related decision, see AI receptionist or answering service for a booking-based business.

Test month: how to try before you commit

Do not sign a long contract on the strength of a demo. Run a structured month.

  1. Week 0: count your calls. Log every call for a week: time, length, who, and what they wanted. This is the data for your sums.
  2. Week 1: pick a plan one size down. Choose the smaller plan that fits your minutes, and ask the vendor how overage is charged.
  3. Weeks 1 to 4: listen. Read the call notes or listen to recordings of at least ten calls. Check that names, numbers and addresses are right.
  4. End of week 4: score it. Count correct messages, missed details, callers who complained, and booked jobs.

Here is an illustrative scorecard for a month on an AI plan.

MeasureResultTarget
Calls answered62 of 62100%
Messages with correct name and number56At least 95%
Callers who asked for a person9Under 10%
Bookings made21Rising month on month
Complaints2Zero to one

In this example the AI missed details on six messages (56 of 62 is about 90 percent), nine callers asked for a person (about 14.5 percent), and there were two complaints. That is not good enough to go solo on every call, but it is good enough to keep for routine calls while transferring enquiries to a person. The test month produces a mixed answer, which is how most do.

One week of calls, sorted and costed

A worked example makes the sums concrete. Picture an insurance broker with five staff who logs one week of inbound calls (illustrative figures, scaled to a month by multiplying by four).

Call typeCalls in the weekAverage minutesMinutes
Policy documents and renewal dates14228
New quote enquiries9436
Claims, upset callers4624
Sales calls and wrong numbers616
Total3394

Over a month that is about 132 calls and 376 minutes. Putting every call behind Ruby would need the 500-minute plan at $1,725, or about $13 a call, because 376 minutes does not fit the 200-minute plan. But 24 monthly calls are sales pitches and wrong numbers that a screen could drop, and 56 are routine document and renewal calls that an AI handles well. That leaves 36 new quote enquiries (about 144 minutes) and 16 claims calls (about 96 minutes). If the staff take the claims calls themselves and only the quote enquiries go to a human service, the human minutes fall to about 144 a month. That needs the 200-minute plan at $720, or $20 a call for 36 calls, which is high per call but each one is a prospective sale. If the quote enquiries also go to an AI with a good script, the human bill could drop to nothing, at the cost of some lost enquiries. The saving here comes from sorting the calls first and avoiding the jump to a bigger plan, not from a headline price.

Before and after: what a good message looks like

The quality of the message a receptionist leaves decides whether you can act on it. Here is an illustrative bad and good version of the same call from a new enquiry.

Weak message:
  "Someone called about a quote, wants a callback."

Good message:
  "New enquiry, 2:40pm Tuesday. Name: [first name], spelled back.
   Number ending 4417, confirmed. Wants a quote for van cover,
   three vehicles, current policy renews on the 14th. Prefers a
   call after 5pm. Urgency: medium. Asked for a person: no."

Check whichever service you choose against that standard during the test month. A human service should do it from the start; an AI service should do it once the script asks for each field in order. If messages arrive looking like the weak version, the plan is not worth the price regardless of who answers.

A short list of questions to put to any vendor

  • What exactly counts as a minute or a call, and is hold or transfer time included?
  • What is the overage rate, and is there a cap so a busy month cannot run away?
  • How long is the contract, how much notice to cancel, and can I keep my number?
  • Who hears the calls, where are recordings stored and for how long?
  • How does a transfer to my staff work out of hours, and who is responsible if it fails?
  • Can I listen to or read a sample of real calls before I sign?

Write the answers next to each vendor in a simple grid. A vendor who answers all six plainly is easier to trust than one with the lowest price and vague terms.

When the cheapest option is no service

Below a certain volume a receptionist service costs more than the leads it saves. If you get 12 calls a month and each lead is worth $200, a $250 plan only breaks even if it rescues more than one missed lead every month. A simple voicemail greeting that promises a call back within two hours, plus a text auto-reply, may do almost as well for nothing, and at this volume a free AI tier (Smith.ai's covers 25 calls a month) costs nothing either. Do the sum honestly: value per rescued enquiry, times the number you would have missed, against the plan price. Many owners find they miss fewer calls than they feared, and they spend the money elsewhere.

Problems to expect with either route

Misheard names and numbers

Both humans and AI mishear. A message that says "call Dave on 555 0142" with a wrong digit costs a lead. Ask for read-back of names and numbers as part of the script, and spot-check messages against recordings each week.

The call that should have been transferred

An AI that keeps asking questions of a caller who is clearly upset does damage. Write transfer triggers: words such as "emergency", "urgent", "complaint", "lawyer" or "cancel", and a limit of two failed attempts before handing over.

A minute plan that never fits

If your calls cluster just above a plan boundary, you pay for unused minutes or overage. Re-check your minutes quarterly and move plan if the pattern changes.

Recording and consent

Call recording and automated voices bring legal duties, which depend on where your callers are. Check the rules that apply to you with a qualified adviser before recording or using an AI voice, and tell callers when a call is recorded or automated.

Is Ruby legit? Checks before you pay

For any phone service, run the same short due-diligence list.

  • Read recent independent reviews on more than one site, and weigh complaints about billing and contracts more than complaints about a single call.
  • Get the terms in writing: minute definition, overage rate, contract length, cancellation notice and what happens to your phone number if you leave.
  • Test with a short plan first, and keep your own number forwarding so you can switch back.
  • Ask where recordings and messages are stored and who can see them, particularly if calls include client details.

I have not tried to rate Ruby against review sites here, since review scores change daily and are easy to manipulate in both directions. The transparency of its pricing page is a good sign; it is not a guarantee about service quality, which only a test month shows.

A decision rule

  • Choose AI if most calls are routine, volume is above about 40 a month, and you can write clear transfer rules. Expect to keep a human fallback.
  • Choose a human service such as Ruby if a large share of calls are sensitive, high-value or emotional, and each lost enquiry costs you more than the plan.
  • Choose a hybrid if calls are mixed. Let AI take the routine ones and route the rest to people.
  • Stay with your own staff, voicemail or a free AI tier if you get fewer than 15 or 20 calls a month; a paid human service rarely pays for itself at that volume.

To check the rule against your own numbers, work through the sum earlier in this tutorial with your own week of calls, and compare it with measuring an AI receptionist's return in 90 days. If you would rather talk the numbers through, that is what my AI implementation consultation covers, and you can reach out here.

Questions before choosing Ruby or an AI receptionist

Is Ruby a real, established service or a scam?

Ruby publishes its plans and prices openly and sells to businesses through a standard sign-up. That is normal for a legitimate service, but publication is not proof of quality. Before paying, read recent reviews on independent sites, ask for the contract terms in writing, and test the service on a short plan.

How is Ruby priced compared with per-call services?

Ruby sells monthly bundles of receptionist minutes: $250 for 50, $395 for 100, $720 for 200 and $1,725 for 500 on its pricing page. Per-call services such as Smith.ai bill by the call. Convert your own calls to minutes or calls before comparing, because a long call costs more on minute plans.

Can an AI receptionist handle emergency or upset callers?

It can take details and route the call, but a distressed or angry caller usually needs a person. Set rules that transfer immediately on certain words, caller types or topics, and check that the transfer works out of hours. Test with a few awkward calls before going live.

Do callers mind talking to an AI?

It depends on the caller and the business. Many people accept a quick AI greeting for a simple booking, but older callers and anyone with a problem tend to prefer a person. Tell callers it is an AI, offer a way to reach a human, and watch complaint rates in the first month.

What does a hybrid set-up look like?

An AI answers first and handles bookings, hours and simple questions, then transfers anything complex to a human service or your own staff. Some human services, Ruby included, list AI-assisted call flows alongside human answering, so ask exactly what the AI does and what is billed.

Further reads

Sources: Ruby pricing page (ruby.com/pricing); Smith.ai receptionist pricing page (smith.ai/pricing) and AI receptionist pricing page (smith.ai/pricing/ai-receptionist). All checked 5 November 2026; check each vendor's page for current prices and terms.

Want to cost your own phone coverage?

On a 1:1 call we can look at your call volume and the calls that really need a person, then decide between a human service, an AI receptionist or a mix.

Book a 1:1 call with me