Cisco gives one engineer unlimited AI tokens to refactor 98% of legacy code

Cisco's president and chief product officer, Jeetu Patel, told Fortune on 30 October 2026 that he has given one engineer unlimited AI tokens to refactor the company's legacy code. The engineer is Padmanabhan Krishnamurthy, 27, who joined Cisco in 2025 when it acquired EzDubs, the real-time translation startup he co-founded. Fortune reports that Patel gave himself four weeks to prove the approach, and a goal of refactoring 98 percent of Cisco's legacy code within 24 months.

What Cisco has said

• Krishnamurthy reckoned he could rewrite the legacy code with a two-person team. Patel told Fortune that even the most AI-forward people on his own team were not thinking that way.
• Patel's words on the budget, as Fortune quotes them: "I've given him unlimited tokens because he's going to refactor code. So I'm like, you can go, you can go nuts."
• Patel is also looking for more young hires he describes as challengers of the status quo, and says friction with long-tenured staff is the point.

What is not known

The free portion of the article does not say what 'legacy code' covers, what languages it is in, how much of it exists, or what 98 percent is measured against. It gives no token spend, no testing or parity method for checking that rewritten code behaves like the original, and no description of what the four-week proof would show. Whether any rewriting has started is also not stated.

That matters because an unlimited budget with a single engineer is a spending cap in practice: one person can only review so much output. A reader should treat this as an announced intention from an executive, not as a deployment with results.

What Hacker News said

The story reached 208 points on Hacker News on 30 October (the host is not on this site's article source list, so the thread is described here but not linked). The strongest objection in the thread is about purpose: one commenter noted the article gives no goal for why 98 percent of the code needs rewriting, and another asked whether one engineer has the domain knowledge to know why the old code was written as it was, or will just reproduce its flaws. A supportive commenter wrote that a similar embedded-system rewrite at their own company took about three months and roughly $20,000 in tokens plus half of a mid-level developer's time; that is an unverified anecdote from the thread, not a Cisco figure.

What a team can take from it

The useful part is the shape of the experiment: a named owner, a short proof window and a stated end goal. The missing parts are the ones to add: a token ceiling or spend alert even if the policy is 'unlimited', a parity test written before the rewrite, and a written stop condition for the four weeks.