Meet Rafi Farouk: From Trading Floors to Legal AI
Rafi's journey is the stuff startup folklore is made of. After studying philosophy and economics at university, he became a trader managing over £1 billion at just 22 years old (yes, including Vatican investments). But craving something with genuine social impact, he pivoted to machine learning and founded Genie AI eight years ago, before GPT-2 even existed.
Key insights from Rafi:
"We're democratizing expertise and what we're really selling is trust."
"The bigger market is all the end businesses, and we feel we're making a bigger difference in the world by serving that mid-market business rather than the big players."
The Technology: Eidetic Intelligence
Genie's secret sauce is a patent-pending architecture called Eidetic Intelligence, which dramatically outperforms ChatGPT and Claude for legal tasks:
86% accuracy on benchmarks vs. GPT's 48% a significant leap in legal quality
Handles unlimited context length across 20-30+ documents simultaneously
Creates a knowledge graph of company policies, templates, and negotiation behaviour
Maintains legal quality through intelligent compression and gating mechanisms
This matters because most law firms operate on a template and tweak model, charging tens of thousands for slightly modified contracts. Genie flips the script entirely.
The Jobs Question (And Why It Matters)
Let's address the elephant in the room: yes, jobs will be lost in law firms. But Rafi isn't dodging the question.
"I do think jobs will be lost in law firms and we shouldn't beat around the bush with that."
However, Genie is partnering with Lord Wahey and Maker Life to create "Trust Engineers"—lawyers reskilled to set AI rules, standards, and guardrails. The company is publishing this training program publicly later this year.
The Bigger Picture: The Agent Economy
Perhaps most fascinating is Rafi's vision for the future of work itself. As AI agents handle more tasks autonomously, company structures are already shifting:
"Everyone is being an autonomous end-to-end team. Everyone's doing the work end to end via agents."
Rather than cross-functional teams slowing things down with collaboration, we're moving towards generalists managing multiple AI agents—essentially, one-person companies at scale.