Felipe Corcuera·September 1, 2025

The Joy & Pain of Buying a Tech Business | Felipe Corcuera Interview

Felipe Corcuera, a Goldman Sachs M&A banker turned search fund investor at Relay Investments, partnered with co-CEO Antonio to run a traditional search fund (Brooklyn Partners) and acquired Beaker, an IT services and robotic process automation (RPA) company based in the Woodlands, TX, in December 2022 for $8.5m (1.6x revenue on $5.5m revenue / ~$1.4m EBITDA), financed with investor equity, a $1.7m seller note, and ~$1.2m equity rollover. Twelve days after closing, ChatGPT launched, threatening the RPA software implementation model that was Beaker's core business. Rather than retreating, Felipe and Antonio pivoted: they built a 'Robots as a Service' recurring revenue model, established an internal AI lab of 6-7 engineers training proprietary LLM modules into plug-and-play AI agents sold on a pay-per-transaction basis, and maintained their profitable staff augmentation business (powered by a university pipeline and Beaker Academy). As of the episode, EBITDA is near breakeven due to heavy reinvestment in AI talent and revenue model transformation, but recurring revenue has grown substantially and ARR from AI agents is approaching $1.5m — potentially valued at 5-7x vs. 1.6x for services revenue. Felipe and Antonio explicitly plan a long-term hold, citing Will Thorndike research showing compounding value peaks in years 6-7.

Deal facts

purchase price: $8.5m
multiple: 1.6x revenue
sde ebitda: EBITDA $1.4m at acquisition
revenue: $5.5m
financing structure: Investor equity + $1.7m seller note + ~$1.2m equity rollover from seller
notes: 140 employees plus 40-50 interns at acquisition. Broker had suggested seller ask for $20-22m; seller accepted ~$8.5m. Deal sourced geographically (Woodlands, TX) via proprietary outreach; taken off-market before broker could run a process. Traditional search fund (Brooklyn Partners) co-founded with Antonio. Two other LOI deals were handed off to other searchers for a 1% cash + 1% equity finders fee each.

Why this business

Growing industry (process automation/RPA expected to grow high-20s% per year this decade), motivated seller with realistic expectations, strong customer logos (PepsiCo, Heineken, Nestle, Aramark), good cultural fit with the seller (also Latino), and steady 25% 3-year CAGR. Felipe and Antonio were not looking for a quick exit and wanted a business they could hold long-term; the company's strong brand in a niche space and its university-pipeline talent model were additional draws.

What's working

Conversion of project-based non-recurring revenue to recurring 'Robots as a Service' (RAS) model with multi-year contracts and automatic renewals, growing ARR meaningfully
Building proprietary AI agents using pre-trained, pre-tested LLM modules ('Lego blocks') sold on a pay-per-transaction model with low setup fees — shifting customers from capex to opex
Staff augmentation business remains organically strong: niche RPA focus, relationships with 15-20 technical universities, internal Beaker Academy trains interns into full-time certified developers, and proactive monthly customer service outreach drives expansion
Blue-chip customer logos (Pepsi, Heineken, Nestle, Aramark) retained for 6-7 years and continuing to consume services, including global teams across LATAM, US, and India
Internal AI lab (6-7 engineers, growing) doing R&D and model training to stay ahead of the market and build defensible AI product IP

What's hard

ChatGPT launched 12 days after acquisition closed (November 30, 2022 vs December 12, 2022 close), creating immediate existential uncertainty around the RPA/automation business model
Revenue transformation from project-based to recurring created severe cash crunch: lower upfront fees + revenue deferred over multi-year contracts + simultaneous investment in AI lab + AI engineers command market-rate salaries
Salespeople were cannibalizing project-based revenue with RAS deals rather than growing the overall top line, masking the underlying shift in revenue quality
Customers demanded AI immediately after ChatGPT, but were not data-ready and didn't know what they wanted — leading to a mismatch between demand and deliverable
Constant threat of disruption: OpenAI or a competitor could release an AI builder tomorrow that eliminates their implementation value; operating on 'mental health on your toes every day'
Bad AI actors in the market (VC-backed startups overpromising) contaminate customer trust in AI, making sales cycles longer even when genuine interest is high
Tried an AI sales agent internally for 6 months and found it underwhelming — essentially just a wrapper around email campaign management

Notable quotes

We bought on December 12th. I think those are the dates. So, we didn't honestly — we didn't make much of it. When it came out, most people, including ourselves, thought, you know, what a funny fun tool to ask for restaurant recommendations.

The broker was encouraging the seller to ask for almost twice as much as what we paid for, which was completely unrealistic. So the seller had done his homework separately and aligned his own expectations despite the broker saying... this is worth 20 or 22 million and the seller was like, no it's not worth 20 million.

Antonio and I are not looking to tell a beautiful story today because we're not looking to sell. So I'd rather reinvest everything and have a low free cash flow number to show for it because I don't need to show that.

One of our board members sometimes says, you should never stop doing things that feel easy. You know, if there's something you're doing that feels like cheating because it's so easy, just do more of that. And staffing sometimes feels like it.