Cisco AI Code Rewrite Led by 27-Year-Old Engineer With Unlimited Tokens

Cisco has put one of its most consequential engineering jobs in the hands of a 27-year-old engineer who, until recently, sat five rungs below the executive suite, launching a Cisco AI code rewrite. President and chief product officer Jeetu Patel personally selected the engineer to lead the major code refactoring effort and gave him an unlimited budget for AI usage, according to Crypto Briefing.

The appointment is one piece of a much larger push inside the networking company. Roughly 32,000 engineers at Cisco are being shifted toward building software in an AI-centric way, and the Cisco AI code rewrite sits at the center of that transition.

The decision to elevate a junior staffer, rather than a senior manager, signals how Patel wants the program to run. He bypassed the layers of leadership between his office and the engineering ranks and handed the assignment directly to the person he wanted doing the work.

Unlimited tokens for the Cisco AI code rewrite

Patel described the move at a small roundtable dinner with Fortune. He said the engineer he chose was positioned five levels below him in the company hierarchy, and that he reached down through the org chart to find him rather than promoting through the usual management channels.

"I've given him unlimited tokens because he's going to refactor code," Patel told Fortune. "So I'm like, you can go, you can go nuts."

In AI systems, text is broken into small units called tokens that models take in and produce, and providers bill customers based on how many of those units are consumed. An uncapped budget therefore removes one of the main constraints on how aggressively an engineer can point AI at a codebase. The assignment itself, refactoring, involves restructuring existing code while leaving its behavior unchanged.

That combination of an open-ended budget and a narrowly defined technical task is unusual inside a large corporation. Most developers working with AI tools operate under usage limits tied to cost controls, which makes the freedom Patel granted this engineer stand out within the Cisco AI code rewrite.

Six AI-written products targeted by end of 2026

The company has laid out concrete targets for the strategy. Cisco's goal is to ship six products written entirely by AI before the close of 2026, and by 2027 it wants AI to generate 70% of its offerings in full, according to Crypto Briefing. Alongside those goals, the company plans to wipe out all of its legacy code over the next two years.

Early results offer some support for the timeline. In one project, engineers used the approach to cut a codebase down from 500,000 lines to 120,000 lines over the course of two weeks, an outcome that helps explain why Patel is willing to stake an uncapped token budget on a single engineer.

That reduction, removing more than three quarters of the lines in the original codebase in a fortnight, is the clearest evidence cited so far that the method can operate at the speed the 2026 and 2027 targets would require. Whether it scales across the full product line is the question the Cisco AI code rewrite now has to answer.

Founder mode inside a decades-old incumbent

Patel has described the strategy in terms of founder mode, running Cisco with the habits of a young startup even though it is an incumbent that has operated for decades. Under that framing, AI takes over the routine portions of coding, and human engineers are left to concentrate on problems that are harder and carry more strategic weight.

The mandate has teeth. Patel has made adoption of AI tools compulsory across the engineering organization, and employees who do not comply face potential risks to their jobs, according to Crypto Briefing.

The founder mode framing also helps explain the choice of leader for the refactoring work. Picking a 27-year-old from five levels down, rather than a veteran executive, fits a model where speed and direct contact with the work matter more than tenure.

Token budgets folded into headcount decisions

The company has begun merging token budgets with the headcount budgets that managers control. Because money for AI usage and money for hiring now come from a single pool, each manager must weigh the choice directly: bring on another engineer, or spend the same funds on additional compute. That structure pushes the philosophy behind the Cisco AI code rewrite into routine resourcing decisions across the company.

The internal push comes as Cisco's external AI business expands. Orders for the company's AI infrastructure business have reached $9.3 billion over the last two years, giving Cisco a commercial stake in the same technology it is now turning on its own code.

Taken together, the budget mechanics, the hiring trade-off and the infrastructure order book show a company aligning its internal engineering practices with the market it sells into. The unlimited tokens granted to one young engineer are the most visible expression of that alignment, but the policy changes underneath it affect every manager in the engineering organization.

Conclusion

Cisco has handed a 27-year-old engineer, plucked from five levels below the executive suite, an unlimited AI token budget to lead a major code refactoring effort, according to Crypto Briefing. The appointment anchors a company-wide shift of roughly 32,000 engineers toward AI-centric software development. The next concrete milestones are the targets Patel has set: six fully AI-written products by the end of 2026, 70% of offerings entirely AI-generated by 2027, and the elimination of all legacy code within two years.

Frequently Asked Questions

Who did Jeetu Patel pick to lead Cisco's code refactoring effort?

Patel, Cisco's president and chief product officer, picked a 27-year-old engineer who sat five levels below the executive suite in the company hierarchy. He described the selection at a small roundtable dinner with Fortune, according to Crypto Briefing.

What does unlimited AI tokens mean for the Cisco engineer?

AI models process language by breaking it into small units called tokens, and vendors charge customers based on how many of those units are consumed. An unlimited budget means the engineer can run AI tools against Cisco's code without a spending cap, a freedom Patel confirmed when he told Fortune the engineer could go nuts.

What are Cisco's AI product goals for 2026 and 2027?

Cisco aims to have six fully AI-written products by the end of 2026, and by 2027 it wants 70% of its offerings to be entirely AI-generated, according to Crypto Briefing. Separately, the company intends to retire every line of its legacy code over a two-year period.

How fast has Cisco's AI refactoring worked so far?

In a single project, the AI-driven approach reduced a codebase from half a million lines down to 120,000 lines over two weeks, according to Crypto Briefing. That outcome is the clearest data point so far suggesting the method can operate quickly.

How is Cisco pushing its 32,000 engineers to adopt AI?

Adoption of AI tools is now compulsory across the engineering organization, and employees who resist could put their jobs at risk, according to Crypto Briefing. Cisco has also folded token budgets into managers' headcount budgets, forcing a direct trade-off between hiring another engineer and buying more compute.