Cisco gave a 27-year-old unlimited AI tokens to rewrite 98% of its legacy code

Cisco president handed a former startup founder unlimited AI tokens to refactor decades of legacy code with just a two-person team.

Business · Source: Notebookcheck

What happened

Cisco is betting big on AI to modernize its massive legacy codebase. President and Chief Product Officer Jeetu Patel handed unlimited AI tokens to Padmanabhan Krishnamurthy. Krishnamurthy is a 27-year-old engineer and former startup founder. He joined Cisco after they acquired his real-time translation startup EzDubs.

Patel found Krishnamurthy buried five levels deep in the organization doing unstrategic work. During a meeting at the San Francisco office, Krishnamurthy pitched a wild idea. He claimed he could rewrite decades of legacy networking code using AI with just a two-person team. Patel told him to go nuts and gave him unlimited token access. Even the most AI-forward people on the executive team had not considered tackling the project with so few people.

The 98 percent rewrite is not a finished result yet. It is a highly ambitious target. Krishnamurthy gave himself four weeks to prove the concept actually works. If successful, the goal is to refactor 98 percent of the legacy code within 24 months. The core engineering team will still take ownership of the final product. Patel wants this tiny group to act as a catalyst for wider change in how Cisco develops software.

Key facts

Why it matters

This proves that massive enterprise refactoring is no longer a headcount problem. A two-person team with unlimited AI tokens can now pitch projects that used to require hundreds of engineers. Cisco wants to modernize its code to support migration to memory-safe languages. This prevents memory errors that lead to critical security flaws. If you are building AI tools, the enterprise market for automated code migration is wide open. But the stakes are incredibly high. Cisco networking devices keep parts of the internet running. Autonomous agents can make hard-to-spot mistakes in these critical systems.

This move also creates deliberate tension between AI-native Gen Z hires and veteran engineers. Patel admits this friction is intentional. He wants younger hires to challenge established approaches. He even pointed to an intern who tackled a redesign that experienced staff had struggled with. The future of enterprise software development will combine experienced operators at scale with AI-native founders who challenge the status quo. Patel calls interns reverse mentors who teach colleagues different ways of thinking.

For builders

Pitch massive refactoring with tiny teams

Enterprise leaders are hungry for AI-driven efficiency. You do not need a massive agency to win enterprise contracts anymore. A small team with the right AI workflow can replace entire legacy modernization departments. Founders who pitch lean AI teams will win big enterprise budgets.

Target memory-safe language migration

Cisco specifically wants to migrate legacy code to memory-safe languages to stop security flaws. Build AI tools that specialize in translating old networking code into modern secure languages. Enterprise security teams will pay massive premiums for tools that automate this transition safely.

Solve the regression testing bottleneck

AI coding agents make hard-to-spot mistakes in critical systems. Cisco software spans thousands of applications and operating systems. The real bottleneck is verifying the work and fixing regressions over the next two years. Builders who create automated AI testing tools will capture massive value.

My take

I love that Cisco bypassed five layers of middle management to back a 27-year-old founder. It shows that in the AI era, clarity and execution matter more than your tenure at a massive corporation. If you have the guts to pitch a two-person team for a massive enterprise rewrite, you can win unlimited resources.

Original reporting: Notebookcheck. This is my rewrite and opinion.

More AI news for builders