Rilla Review: The AI That Rides Along to Every Sales Call You Never See

Somewhere right now, a sales rep is standing in a stranger's kitchen quoting a new HVAC system. No manager is there. No recording is running. The rep will come back with a signed contract or a shrug, and that outcome is the entire performance review.

Now multiply that by forty reps. Two of them close six appointments out of ten. Two close two out of ten. That gap is worth hundreds of thousands of dollars a year, and nobody at the company can explain it, because nobody has ever heard those conversations.

The traditional fix is the ride-along, where a manager tags along to watch. A manager can realistically do two a week while the team runs fifty appointments. That is about four percent visibility, which is a bit like coaching a football team by watching two plays a season and guessing about the rest.

So what is Rilla?

Rilla is a mobile app that records in-person sales conversations, then uses AI to transcribe, summarise, score and analyse them. The rep taps record walking up the driveway. The manager reviews it later, at speed, from a desk or from a truck between appointments.

The company calls this a "virtual ridealong," and the analogy underneath it is film study. Athletes have had game tape for decades: every play broken down, every habit named. Salespeople never had tape. They had memory, and memory is extremely generous to whoever is doing the remembering.

The part that matters more than the recording, though, is what happens next. Rilla says that if a company has no script, its AI learns from the top performers and coaches everyone else toward that standard. The real output is not audio files. It is a working definition of "good," extracted from the people in the building who are already good, then applied to everybody else.

Customer numbers are specific enough to be checkable. Franchise network Neighborly ran 5,000 virtual ride-alongs in thirty days across 130 technicians. A sales manager at Cardinal HVAC went from six or seven ride-alongs a week to twenty-five or thirty. Rilla's own marketing claims an average 40 percent lift in close rates and 17 percent in ticket price, which are vendor figures and should be read as such. The logo wall is more persuasive than the percentages: Home Depot, Kohler, Midas, Neighborly, Groundworks and ARS.

The part nobody expects

Rilla did not arrive fully formed. Founded in 2019 in New York by Sebastian Jimenez, Christopher Martin and Michael Castellanos, it spent years losing money and hopping between markets looking for someone who cared. Then it landed on field sales and hit $130,000 in revenue in its first month in the new vertical.

The most interesting number in the company's history is the price. Rilla went from charging about $30 a month to roughly $5,000 a year per licence. The technology did not get 150 times better. It found customers whose problem was expensive enough to justify the bill, which is the quiet lesson buried in this product: the software changed far less than the person paying for it did.

Investors noticed. Rilla has raised roughly $78.9 million, with a Series B led by GV in June 2025 and a valuation reported around $737 million. Research firm Sacra estimates it reached about $70 million in annual recurring revenue by April 2026, an outside estimate rather than a company disclosure.

What is actually in the box

• Recording and transcription tuned for face-to-face audio. Most speech models were trained on call-centre and video-meeting sound, which is a different problem from a phone in a shirt pocket in a noisy basement.
• Virtual ride-alongs, where a manager skims, searches and jumps straight to the moment the price came up.
• Scorecards that check each appointment against the company's playbook and flag where the rep went off-script.
• Behavioural analytics covering talk-time ratio, patience and interactivity. Rilla's own research puts top performers at 45 to 65 percent of the talking, which is a useful thing to be able to measure at all.
• Rilla Live, which lets a manager listen to an appointment in progress and send notes.
• Roleplay, where new reps practise against a bot built from real objections instead of practising on real customers.
• Rilla Intelligence, an aggregated layer sitting on metadata from more than 1,300 companies. Rilla has used it to publish cross-client studies, including one on the best days and times to book an in-home consultation.

Where it falls down

The microphone is the weak link. Reviewers describe muffled pocket audio, transcription errors around strong accents and background noise, recordings that stop mid-presentation, and appointments occasionally filed under the wrong rep. Some report grading consultations by hand because the AI scorecards were not accurate enough to trust. The analysis is only ever as good as what the phone heard through a jacket.

The setup claim deserves a raised eyebrow. Rilla's FAQ says onboarding takes one hour and forty-five minutes. G2's aggregated buyer data reports about a month to implement and eight to ten months to see a return. Both can be true, because switching software on is not the same job as changing how forty people work.

Adoption is the real risk. Recording is a habit, not a feature, and habits decay. Competitor Siro markets against Rilla on exactly this point, claiming licences that go unused, which is a competitor claim and should be weighed accordingly. The underlying concern still holds: a rep who forgets to press record produces nothing to coach.

The contract is rigid. Annual, paid upfront, auto-renewing. One G2 reviewer describes a $13,000 renewal charge they say arrived without notice and could not be refunded. That is one account rather than a pattern, but the structure makes it possible, so the renewal date belongs in a calendar reminder.

Read the reviews with one eye open. A large share of Rilla's Capterra reviews carry a label noting the reviewer was invited by the vendor and offered an incentive. The scores are strong, and that context still matters.

It covers one leg of the journey. Rilla coaches the in-person appointment. It does not coach the call-centre agent who booked it or automate the follow-up that rescues it.

What it costs

Rilla does not publish pricing, which is itself informative. The figures below are triangulated from independent research and the founder's own public comments, and every one of them should be confirmed against a real quote.

Search "Rilla pricing" and the results are almost entirely pages published by companies selling alternatives, with per-user estimates ranging from $80 to $500 a month. That is a spread of more than six times, from sources with an obvious interest in the answer. Treat the whole genre as noise and get a quote.

The break-even test is simple enough to do on a napkin. At the reported floor, five reps cost roughly $20,000 a year, so each one needs to produce about $4,000 in additional gross profit. For a team selling five-figure jobs in people's homes, that is one or two extra wins. For a team where face-to-face selling is occasional, the maths stops working quickly.

Privacy and the fine print, in plain English

This product records private conversations inside customers' homes, so this section is not optional reading.

Rilla's public FAQ answers the legality question with a flat yes, then names eleven US states where the customer must be told first. Its privacy policy, effective 31 July 2026, takes a different tone: it positions Rilla as a service provider and puts responsibility for giving required notices and obtaining consent on the customer, meaning the contractor, not Rilla. Read those two documents together and the message is that the marketing says it is fine while the contract says it is your problem. Anyone deploying this should get the disclosure script cleared by their own lawyer rather than by a vendor FAQ.

Three other things worth knowing. Audio, transcripts and related usage data may be sent to a third-party AI provider for processing. Everything is hosted and processed in the United States. And Rilla reserves the right to create aggregated, de-identified data from what it collects, use it to improve the service, and share it with third parties, which is standard language that reads differently when the raw material is recordings of people's living rooms.

One flag for procurement: at the time of writing, Rilla's public security page contained placeholder text, and no published security certifications could be found. That is not evidence of weak security, but it does mean the documentation has to be requested directly during the sales process rather than verified in advance.

The verdict

Rilla answers one question extremely well, and it happens to be a question that has gone unanswered for the entire history of field sales: what is actually being said in the appointments nobody attends. Inside that lane the product is strong, the customer results are specific rather than vague, and the coaching layer solves a genuine management problem instead of an invented one.

Outside that lane, the case narrows fast. It is field-only, priced for teams that live on the doorstep, locked to annual contracts, opaque about cost, dependent on reps remembering to press record, and limited by what a phone in a pocket can hear.

It does not make anyone a better closer. It ends the guessing about why they are not.

Verdict: 3.5 / 5

Buy it if a team runs high-ticket in-person appointments at volume and management is flying blind. Skip it ifface-to-face selling is a side channel, the budget cannot absorb a five-figure annual commitment upfront, or the coaching gap sits in the call centre rather than the field.