Can legacy services or data-rich businesses be converted into great AI companies?

Gokul Rajaram · published Jun 25, 2019

Building a fully verticalized, traditional-looking business is an effective way to collect data in the format you actually need and then automate parts of it with AI.

Data alone is insufficient if it isn't in the needed format; many companies struggle even after getting data, and a verticalized business leaves a viable low-margin fallback.

source Death of a startup

Maor Shlomo · published Nov 24, 2025

The bigger opportunity is building or buying the real-world business end to end — a law firm, a hospital, a bank — and making it AI-native, rather than selling software to it

Model providers and normal software entrepreneurs will not do it, so it is structurally defensible, and the platform shift makes previously bad ideas viable

Scope: contrasted with building a competitor to Harvey or Spellbook; payoff framed on a two-to-four-year horizon

43:14 20VC: Base44's Maor Shlomo on How Vibe Coding Will Kill SaaS and Salesforce | Why it is BS that Vibe Coding Platforms Do Not Have Defensibility and Bad Margins | Why He Worries About Google, Not Replit and Lovable | Why Long Anthropic, Not OpenAI?

Alex Rampell · published Jan 12, 2026 · hedged

Private-equitizing venture capital — buying companies and adding AI to them — is less attractive than he originally argued, because of founder-market mismatch

It feels like a founder market mismatch

67:42 Founder market mismatch makes buy and ai ify unattractive

Harry Stebbings · published Jun 16, 2025

There are far more 'zombies' than 'vampires' because retrofitting AI into an existing company is a much harder transition than it looks

Real product change comes bottom-up, so you can't just slap AI onto an existing product

76:21 Retrofitting ai into legacy companies is harder than expected so zombies outnumber vampires

Brendan Foody · published Jun 1, 2026

AI-enabled services is a genuinely attractive category, but only for companies that derive real competitive advantage from AI rather than merely bolting AI onto a services business

Many companies build services without gaining meaningful competitive advantage from AI; the value comes from actually leveraging AI to automate the delivery work

40:55 Only services with genuine ai derived advantage qualify

Julien Bek · published Aug 24, 2026

He would not invest in a services business planning to convert into a software business; such companies should be built from first principles instead

The best companies concentrate talent and frontier talent won't join a legacy services business dressed up as an AI company; and having data isn't the same as having the right data, which is hard to harness

64:47 Build from first principles frontier talent wont join legacy

Julien Bek · published Aug 24, 2026

Traditional private equity may do very well buying data-rich businesses in the AI era, but is unlikely to produce the next trillion dollar company

Those companies have the data but can't hire frontier talent, and PE is seeking different outcomes than venture is

66:21 Good pe returns but never the trillion dollar outcome