The Trillion-Dollar Industries AI Is Disrupting: Voice, Law & the End of the Billable Hour

The conversation opens with the company's growth figures. One participant notes outside estimates of 350 million in revenue within two or three years, and says they are now hearing "five or 600 million." The company's own account starts in 2022, which it describes as a year spent on research and product. In early 2023 it released what it calls the first text-to-speech model "that finally could sound human."

The stated milestones after launch: - About 20 months to reach the first $100 million in ARR. - About 10 more months to reach $200 million. - About 5 more months to reach $300 million, which is where the company says it closed the end of last year. - "Now we are at $600 million in revenue."

These figures come from the company itself in conversation. Nothing in this part of the interview verifies them independently. The interview also moves between ARR and "revenue" without defining either term.

Asked how it keeps its culture while investors push money at it, the company gives a headcount of 600 people. It describes itself as combining research and product to build "a communication platform for AI." On the research side, that covers generating speech, transcribing speech and orchestrating speech for interactions. On the product side, it covers the customer journey from marketing assets and international localization to voice-agent customer support and proactive help in operations, training and sales. The company presents part of its revenue growth as a result of adding these functions over time.

On retention, the company says its first 10 people, who were mostly research and engineering staff, have had zero attrition. It credits a co-founder described as an "incredible researcher" for assembling a team motivated by audio and interaction problems. It also notes that starting in 2022, when "topics of the day were crypto and Metaverse," gave it time to focus on what it believed was the future.

The discussion turns to how AI coding tools have changed software building, and how quality holds up when non-developers ship code. The company describes an organization built from small teams of five to 10 people. Some are organized around product engineering. Others are go-to-market teams aimed at specific industries such as telco, financial services and healthcare.

Its distinctive choice is to put engineers inside every team, including non-engineering ones: talent, legal, and revenue or go-to-market engineering. These engineers have two jobs. They build automations for their team, and they help colleagues adopt AI while checking the security of everything those colleagues deploy. The stated reason is that people new to building software can often create things but cannot always review what that software does behind the scenes.

if you are not using a lot of the coding software, a lot of the co-working software, then you are probably in the wrong spot. If you're using too much of it, that is also a flag because you maybe are not doing that in the right way.

The interviewer side adds two risks of everyone building software: a leak once the software is in production, and tools left to decay after the person who built them leaves the company. The interview raises these risks but does not explain how the embedded-engineer model handles the second one.