A Product Market Fit Show | Startup Podcast for Founders · Mistral.vc

He fired all his customers. Then built a $1B startup in 2 years. | Jay Madheswaran, Co-Founder of Eve

·52 min·3 clips

Jay says the new product drove 40% cold-outreach conversion and 90% demo-to-pilot conversion.

1. A Product Market Fit Show | Startup Podcast for Founders follows Jay Madheswaran as he explains EVE’s move from horizontal NLP tools to legal AI for plaintiffs. 2. Jay Madheswaran is the co-founder of EVE, and Pablo is the host who presses him on the pivot, the funding, and the product-market-fit evidence. 3. The episode asks what changed between EVE’s early horizontal product and the plaintiffs-focused product that later raised $100 million at a $1 billion valuation. 4. Jay says he was an engineer at Facebook before the IPO, worked at Rubrik, and later joined Lightspeed as an early-stage AI investor. 5. He describes 2018 as a period when many companies called themselves AI companies, but the models were not yet strong enough for useful classification or extraction tasks. 6. He points to BERT, transformer architecture, and transfer learning as the shift that made entity extraction on contracts and bank statements materially better. 7. Jay says the original company idea in 2020 was task automation for NLP workflows, similar to early RPA bots but focused on text-heavy operations. 8. He says EVE was deployed in legal, finance, and other Fortune 500 settings before the product narrowed. 9. He explains that the company raised $4 million in the week of COVID, and that paperwork delays even affected payroll and EIN processing. 10. He says the legal use case became high-volume document processing, especially contracts, medical records, and bankruptcy or collections workflows with many document variants. 11. Jay distinguishes plaintiffs’ firms from big law firms by their contingency-fee model, where lawyers get paid when they win a case rather than by the hour. 12. He says the first legal workflow was case evaluation, where a firm decides whether to take a new caller as a client. 13. Jay says the first product after the GPT era was a transcript-based phone-call system, which he describes as the first product that felt mind-blowing to many buyers. 14. He says that in late 2022 the team was still a horizontal machine-learning startup with about nine employees and roughly $2 million to $3 million in revenue. 15. He says the hard business question was whether to keep the broad model or raise ACV from about $6k to $60k in order to keep growing. 16. Jay says customer discovery involved asking firms what they would drop everything to buy that quarter, then iterating week by week on the answer. 17. He says the internal tension lasted for about three months in early 2023 while the team still supported old customers and built the new pipeline. 18. Jay says the strongest signals were 40% conversion from cold outreach to demo requests and 90% demo-to-pilot conversion on the new legal workflow. 19. The episode has a founder-interview pace, with Pablo pushing for numbers, timelines, and decision logic while Jay answers in a detailed, reflective style. 20. Listeners who like founder pivots, AI product strategy, and legal-tech specifics should listen, while people wanting a light, nontechnical conversation may skip it.

As heard by us

A blunt look at when customer resistance becomes the signal.

One small caveat: the middle moves fast. Still, this is a practical look at a founder deciding to shut down other business lines and put everything into one product while the product was still barely there. What lands is the behavior, not the pitch.

Read the full review in PlayNext →

Why you'd press play

You want a blunt founder story about shutting down old business lines while the old product still had customers.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from A Product Market Fit Show | Startup Podcast for Founders