"Your customer's voice, everywhere you make decisions."
New York, 2016. Three founders, one hunch: the most valuable thing a company owns is buried in the calls nobody re-listens to.
A lender picks up. A patient calls a clinic. A homeowner books a pest inspection. The conversation lasts nine minutes, ends politely, and vanishes - the way roughly 98% of business calls always have. A human quality team might listen to one in fifty, if it has the stomach for it. The rest evaporate.
VoiceOps exists to make that disappearance stop. The company builds AI that treats every conversation - call, email, SMS, chat - as structured data waiting to be read. It listens for intent, objections, promises, and outcomes, then grades the call, flags anything a compliance officer would wince at, and quietly writes the coaching note a busy manager never had time to. It does this for the fiftieth call the same way it did the first. No sampling. No backlog. No "we'll circle back."
That is a mildly unglamorous mission - "we read your phone calls" does not exactly sparkle at a dinner party. But it turns out the unglamorous stuff is where the money and the risk live.
In 2016, Ethan Barhydt had a specific frustration. He had produced educational content at General Assembly and watched, up close, how badly call-center teams were coached - by gut, by anecdote, by whoever shouted loudest in the Monday meeting.
He teamed up with Nathaniel Becker, a data scientist out of LinkedIn, and Daria Rose Evdokimova. Their pitch was blunt and a little cheeky: be the Moneyball of sales. Stop coaching reps on hunches. Measure what the best ones actually say, and teach everyone else to say it. Y Combinator bought in. So, later, did Accel, Founders Fund, and Bain Capital Ventures.
The first product was a Coaching Enablement Platform: analyze the sales and collections calls, hand managers the receipts, make coaches dramatically more effective. It worked well enough that Barhydt landed on the 2018 Forbes 30 Under 30 list, and well enough that customers started reporting close-rate improvements of up to 25% within a month.
But coaching was only the doorway. Once you can read one call perfectly, the obvious question is: why not read all of them, in every channel, for every reason a business might care?
VoiceOps learns from your conversations directly - identifying products, intents, and outcomes without a human babysitting a library of prompts. Here is where that lands day to day.
Surfaces what top reps actually do on winning calls, then turns it into coaching notes and playbooks the rest of the floor can copy.
Auto-fills custom scorecards across every call, learns from your QA team's overrides, and retires the sampling spreadsheet for good.
Flags violations and brand missteps the instant they happen - built for regulated worlds like lending, healthcare, and education.
Ingests calls, email, SMS, and chat in 120+ languages, then pipes intents, objections, and churn signals into your CRM and BI tools.
Real conversations are a mess. People talk over each other. They mumble. They wander off to ask about the weather before circling back to the actual objection. Accents vary. Somebody's dog barks.
Most transcription tools treat that noise as an error to be smoothed away. VoiceOps treats it as the point. Its models are built to parse interruptions, accents, and off-topic side comments - the human texture that a literal transcript flattens into nonsense. The system also redacts PII on the way through, so reading everything does not mean storing everything a lawyer would fear.
The result is a read of the call closer to what your sharpest manager would catch than to what a keyword search returns. That is a harder problem than it sounds, and it is the reason customers keep using the word "accuracy" when they talk about the product.
Compare that to the field. Gong and Chorus point their conversation intelligence at enterprise sales pipelines. Balto works in real time on the live call. VoiceOps aims squarely at high-volume B2C operations where QA and compliance - not deal forecasting - are the thing keeping leaders up at night.
The core pitch, in one picture: manual QA can only ever sample. Software reads the whole book.
Figures reflect VoiceOps' published claims and customer reports; individual results vary.
No unicorn logos for their own sake here. VoiceOps serves the businesses that live and die on the phone - lenders, insurers, home services, healthcare, education. The unglamorous middle of the economy.
Ethan Barhydt, Nathaniel Becker, and Daria Rose Evdokimova start VoiceOps to make sales coaching a measured discipline.
Early backing from Accel, Founders Fund, and Y Combinator (~$2.1M) gets the coaching platform off the ground.
Accel and Y Combinator return to scale coaching for call-center reps. TechCrunch takes note.
Co-led by Bonfire Ventures and Twelve Below, with Precursor, Not Boring Capital, Village Global, and angel Varsha Rao. The platform expands beyond calls into a full multi-channel intelligence agent.
Return to that ringing phone. The lender, the patient, the homeowner. Nine minutes, a polite goodbye - and then, in the old world, silence.
In the world VoiceOps is building, the call does not vanish. By the time the rep sets down the headset, the conversation has already been read: scored against the scorecard, checked for the disclosure that regulators require, mined for the objection that a dozen other callers raised this week. The manager gets a coaching note she did not have to write. The compliance officer gets an alert instead of a lawsuit. The strategy team gets a signal instead of a shrug.
That is the whole trick. VoiceOps did not invent the phone call. It just refused to let it disappear - and turned the most-ignored data a company owns into something it can finally act on. Listen at scale, learn instantly, adapt overnight. Somewhere, right now, another phone is ringing. This time, someone - something - is actually listening.