Inc. contributor and Babson College professor Peter Cohan profiles two of Entrée Capital's most compelling generative AI portfolio companies — one of which is Darrow — using them as a lens to examine what business leaders can learn about where generative AI creates real, defensible value.
Cohan opens with the broader investment context: venture capitalists poured nearly $50 billion into generative AI in 2023, and the companies delivering the strongest returns are those applying the technology to boost productivity in specific, high-stakes industries rather than competing against foundation model platforms. He interviews Ran Achituv, managing partner at Tel Aviv-based Entrée Capital, who explains the firm's thesis: invest in generative AI applications replacing real people doing real work, target a $1 billion exit, and stay away from the platform layer.
Darrow is presented as a prime example of this thesis in action. Cohan describes it as a GPT that mines data for legal cases — one operating without hallucinations in a closed domain, processing tens of millions of files with specific use cases. Achituv's assessment is direct: the company is doing very well financially, growing rapidly, and operating in a domain where AI can deliver precision because the parameters are well-defined and the data is structured around specific legal frameworks.
The piece frames Darrow alongside SWAAP, a construction planning platform, as evidence that the highest-value generative AI companies are those solving concrete, data-intensive problems in industries with historically poor technology adoption. Both companies replace laborious manual work with AI-driven analysis — and both are growing because they deliver measurable return to their customers rather than incremental efficiency gains.
For business leaders reading Inc., the takeaway is clear: the generative AI opportunity lies not in building the next ChatGPT, but in finding industries where expert work is bottlenecked by the volume and complexity of data, and building AI that unlocks that work at scale.
