Pitch, Build, Scale

The episode contains useful domain-specific insights about deposition modernization, civil procedure rules, and the market dynamics of legal tech, but is padded with considerable throat-clearing, personal anecdotes (80-year-old stepfather using ChatGPT), and standard entrepreneurship platitudes. The substantive content - rules admissibility requirements, the 99.6% accuracy threshold vs. 95% court reporter standard, the shift from 10% to 90% video adoption post-COVID - is real but not densely packed.

“Prior to Covid, less than 10% of these depositions use video. So 90% of these depositions were a stack of paper with words on paper.”

“you have the video, that's the official record, you have the human proofread transcript that's like in the 99% uh, accuracy range”

The core insight - applying modern video and AI to replace aging stenographic court reporting - is sensible but not particularly original or contrarian. The framing around the pandemic-driven shift from in-person to remote depositions is somewhat fresh, and the observation about rules being written in the 1990s is valid. However, the episode largely repeats familiar disruption narratives and the AI-as-tool-not-replacement talking point is well-trodden.

“it became really hideous to think about, hey, I'm going to go ahead and invite somebody with a funding machine to type the spoken word”

“the other thing I was going to mention real quickly is the other thing that, that that's lagging in this space are the rules, the rules associated with those civil procedure rules I mentioned earlier. Most of them were written in the early 90s”

Tom Irby is a strong fit: 35 years of executive experience, founder of a court reporting firm he scaled and sold to Veritext (the largest in the industry), and now co-founder of an active legal tech startup doing ~$X million ARR after 24 months. He has direct operator credibility in the space he's disrupting. However, he is not household-name caliber and the episode does not establish revenue, customer count, or funding details that would signal exceptional scale.

“Tom has over 35 years of executive experience and has built, scaled and exited multiple businesses, including a court reporting firm. He sold to Veritext, the largest in the industry.”

“we formed the company in 2022, we, we launched in mid 2023. So we're, we're right at a 24 month window of, of being commercial”

The episode includes specific data points: 25+ states operational, 99.6% accuracy vs. 95% court reporter standard, 5-7 day human proofreading timeline, 90% of post-COVID depositions now use video vs. <10% pre-COVID, 40% shortage of court reporters nationwide, California court system crisis declared. However, it lacks concrete customer names, revenue figures, pricing models, case studies with metrics, and specific competitor positioning. The guest speaks in generalities about 'biggest bucket of litigation' and 'bread and butter market' without numbers.

“The last number I have is 25 and that's probably a month old number, so I'd say it's probably still under 30”

“you have the video, that's the official record, you have the human proofread transcript that's like in the 99% uh, accuracy range”

The host asks reasonable setup questions but rarely pushes back or follow up rigorously. The 'magic wand' closer is soft. The host does probe on regulatory barriers and transcript accuracy handling, which are good questions, but rarely challenges claims or digs into contradictions (e.g., if 99.6% accuracy is so much better than 95%, why is adoption so slow?). The conversation reads more like a friendly interview than a substantive interrogation. Host also allows long wandering answers without redirecting.

“What are the legal or regulatory barriers that go into making this kind of thing?”

“Do you have concerns and how do you address the concerns out there, um, around AI seemingly replacing human jobs, human things that humans do?”