Felipe Corcuera bought an $8.5M RPA shop 12 days after ChatGPT launched

The Setup Felipe Corcuera grew up in Mexico City, did engineering, spent 4.5 years at Goldman Sachs on M&A, then MIT Sloan in 2016. He stayed in the search fund world as an investor at Relay Investments for three and a half years before raising his own traditional search fund with Antonio Elosua in 2021. The partnership ran an unusual structure. Instead of two parallel funnels, they built one. Antonio owned top-of-funnel through NDA. Felipe owned NDA through closing. Over 12 months they ran 155 email campaigns (154 industry-focused, 1 geographic), touching roughly 640 new companies per week. NDA conversion landed around 20% from intro calls. The irony: 154 industry campaigns and the deal came from the one geographic campaign targeting The Woodlands, Texas. The seller was Latino, had just signed with a broker but had not released teasers yet. Felipe and Antonio learned his wish list, structured around it, and got the LOI signed before a process ever started. The Deal Beecker was a Robotic Process Automation consulting firm with 140 full-time staff plus 40-50 interns. Revenue of $5.5M, EBITDA of $1.4M, three-year CAGR of 25%. Client list included PepsiCo, Heineken, Nestle, and Aramark. Revenue mix was roughly 50% staff augmentation, 30% non-recurring bot development, 20% UiPath license resale. Deal terms at $8.5M EV, which worked out to 1.6x revenue: - Seller note: $1.7M - Seller equity rollover: roughly $1.2M - Investor equity for the balance - Traditional search fund structure Close date: December 12, 2022. ChatGPT had launched November 30. Felipe's initial read was the same as everyone else's: cute consumer tool, not a threat to enterprise RPA. Operating Moves Two pivots in under three years. First move, pre-AI-panic: convert project work to recurring. They launched Robots as a Service (RaaS), trading upfront implementation fees for subscription contracts. This squeezed near-term cash flow but started building ARR. Second move, post-AI-panic: stand up an internal AI lab with 6-7 engineers and build pre-trained AI agents that deploy on a pay-per-transaction model instead of project work. The thesis: agentic AI was going to kill traditional RPA consulting, and the only defense was to become the thing that killed it. Revenue mix today: - 40% staff augmentation - 40% recurring (RaaS plus AI agents) - 15% license resale - 7-8% non-recurring projects The valuation math behind the pivot is blunt. AI-native revenue trades at 5-7x. Services trade at 2.5-3x. Moving the mix is worth multiple turns even at flat dollars. Where They Are Now EBITDA is roughly break-even after R&D reinvestment. Felipe is explicit that he could turn it back on: 'I could technically stop investing in R&D and my AI team and the EBITDA would go up like crazy, but that would hurt the company in two or three years.' The business is in the J-curve on purpose. Enterprise clients from the old book (PepsiCo, Heineken, Nestle) are funding the transition. Unlike most searchers targeting a 5-7 year exit, Felipe and Antonio are holding long. The economics of a services-to-software...