Cisco gives engineer unlimited AI tokens to rewrite legacy code
Cisco's President and CPO has tasked a 27-year-old engineer with
The Lowdown
Cisco is embarking on an audacious project to overhaul 98% of its legacy codebase within 24 months, using AI to drive the transformation. This monumental task has been entrusted to Padmanabhan "Paddy" Krishnamurthy, a 27-year-old former startup founder, who has been granted "unlimited AI tokens" by President and CPO Jeetu Patel.
- The Unexpected Prodigy: Paddy, whose previous startup EzDubs (a real-time translation service) was acquired by Cisco in 2025, was found "buried away in the organization doing something unstrategic" before Patel recognized his potential.
- The Mission: Paddy has been given an initial four weeks to demonstrate the viability of using AI for refactoring, with the ultimate goal of rewriting nearly all of Cisco's legacy code. He believes he can achieve this with just a two-person team.
- Generational Vision: Patel champions the hiring of young, "AI-native" talent, seeing their fresh perspectives as crucial for challenging the status quo and injecting new ideas into the company, even if it creates internal "tension."
- Wider Trend: This approach aligns with other tech leaders, such as OpenAI investor Josh Kushner and artist Will.i.am, who actively seek out young,
The Gossip
The Cynics' Code: Doubts and Disasters
Many commenters express deep skepticism regarding the feasibility and motivations behind Cisco's ambitious AI-driven rewrite. They question the lack of clear goals, the potential for introducing new bugs without sufficient testing (especially given the complexity of Cisco's real-world network products), and whether the project is more about PR and buzzwords ("unlimited AI tokens") than a genuine technical strategy. Some view the spotlight on the 27-year-old engineer with suspicion, seeing it as a setup for blame or CEO hero-worship rather than a sound technical initiative.
AI's Ascendancy: Rewriting the Rulebook
A contrasting perspective argues that AI fundamentally alters the traditional risks associated with code rewrites. Proponents suggest that advanced AI models can effectively handle documentation, generate comprehensive test suites (even to 100% coverage), maintain API contracts, and efficiently identify and fix bugs, drastically reducing developer time. For technical debt-laden internal systems, AI-assisted rewrites are seen as a way to achieve significant forward momentum where human teams have historically been stuck, potentially offering a 10x or even 100x acceleration in development.
Cisco's Code: Past Glories, Present Woes
Commenters reflect on Cisco's evolution, noting a perceived decline from its 90s dominance to a current state of "messy" and "bloated" products. Many agree that Cisco's legacy code is likely a significant problem, citing examples of product issues like "hot bricks" and difficult-to-manage features. This context leads some to conclude that while the AI rewrite is risky, Cisco might have little to lose given its current reputation and the urgent need to modernize its offerings to compete with newer rivals like Juniper/Mist and Arista.
Generational Code: The AI Native's Edge
The article's emphasis on a young, "AI-native" engineer sparks discussion about the relevance of age and experience in the AI era. Some question the 27-year-old's specific qualifications for such a massive infrastructure rewrite, given his background in a translation startup. The term "AI-native" itself is debated, with some commenters finding it a valid distinction for those who instinctively grasp AI's potential, while others dismiss it as "AI naive" or a marketing ploy. There are also critical remarks about the perception of Gen Z's work ethic and an unfortunate, albeit brief, discussion on racial stereotypes concerning Indian developers and their approach to code quality and AI adoption.