Most companies founded before the generative AI wave are now, by their own structure, obsolete. That was the blunt argument Gigi Levy-Weiss, co-founding partner at venture firm NFX, made to a room of executives at EY Israel's JOURNEY 2026 conference at the Hilton Tel Aviv on Tuesday. His question to the audience was simple and uncomfortable: if you started your company today, would it look anything like it does now?
A Call for Structural Reinvention, Not Upgrades
Levy-Weiss described what NFX calls "the second founding" - a deliberate break from incremental adoption of new tools toward a full redesign of how a company operates. Installing ChatGPT or Claude into existing workflows, he said, is "table stakes," not transformation. Speeding up individual tasks with AI assistance does not make a company AI-native. The distinction he offered as a test: has the actual path from identifying a customer's need to delivering a solution changed, fundamentally, in shape and speed? For the overwhelming majority of businesses, he argued, the honest answer is no - because the underlying organization hasn't changed, only the tools layered on top of it.
The Org Chart as the Obstacle
The core of his argument targets corporate structure itself. Departments and specialized roles made sense, he said, when expertise was narrow and scarce, and coordination between specialists was unavoidable. But every handoff - from product to marketing to engineering - introduces loss: information degrades, context disappears, speed suffers. His proposed alternative is the "pod": a small, cross-functional team holding full ownership from need discovery through delivery, operating without the friction of departmental boundaries. This is newly viable, he argued, because individual employees have shifted from executing narrow tasks to orchestrating AI systems that do much of the execution themselves.
Technology First, People Second
That shift reorders how problems get approached inside a company. Rather than starting with "who should do this," Levy-Weiss said the first question now has to be whether a task can be automated through AI or autonomous agents - only afterward does the question of which humans are needed to run and govern that automation arise. Speed, too, gets redefined: not the pace of internal processes, but the pace of actually getting a working solution into a customer's hands. Faster cycles allow more experiments, and more experiments mean more chances for something to succeed.
An Uncomfortable Verdict on People - and Leaders
The talk's sharpest edge concerned workforce composition. Levy-Weiss argued every role, including finance and marketing, must become a "builder" role, and acknowledged without softening it that many current employees may not be suited to the organizations companies need to become. He raised the ethical weight of that statement and then set it aside rather than resolving it. He extended the same scrutiny to leadership, including himself, suggesting founders and executives often become limitations once their ambitions are shaped by what was previously possible rather than what AI now makes achievable. His closing instruction was unambiguous: build a company capable of outgrowing its own founders, because if you don't, a competitor will.