Running an experiment and deploying its output are separate decisions. We can explore what is possible, assess the quality, and understand the economics before deciding what belongs in production. We won’t know how far this approach can take us until we try.
Jeetu PatelPresident and Chief Product Officer · Executive, 
Cisco
Social post: Jeetu Patel's LinkedIn post about modernizing Cisco's code, LinkedIn, November 2, 2026.
Jeetu Patel, President and Chief Product Officer. Runs the Cisco organization whose experiment he describes, so an insider; the experiment's results are not yet reported.
Cisco sells AI or agent tools or services, so it has a reason to publish flattering numbers about AI. The passage stays in the archive; read it with that in mind.
Checked against the source by a first reader, then confirmed independently by a second reader who fetched the source without seeing the first reader's notes (Independent review agent (Claude Sonnet), blind second pass from passage text and citation only, no repository access, 2026-11-04).
Found in Patel's post. The text matches exactly, ending "until we try."
Jeetu Patel (Cisco), “Jeetu Patel's LinkedIn post about modernizing Cisco's code”, LinkedIn, 2026-11-02. https://www.linkedin.com/posts/jeetupatel_lot-of-discussion-on-a-fortune-article-link-activity-7523171493376090112-C67s Archived at https://the-ai-record.com/passages/cisco-running-experiment-deploying-its-output-separate-2026
The idea is to combine rigorous simulations and evals, with swarms of AI agents to rewrite a piece of existing code in a memory-safe language. We want to see how quickly we can migrate it, make the code more compact, and create a better foundation for shipping new features.
Jeetu PatelPresident and Chief Product Officer · Executive, Cisco
Social post: Jeetu Patel's LinkedIn post about modernizing Cisco's code, LinkedIn, November 2, 2026.
The hardest and most consequential work are the simulations and the evals in addition to the test pipeline. We need to test whether the rewritten software preserves the behavior customers depend on, performs under real conditions, and handles failures correctly. The agents can iterate, but passing those comprehensive evals has to be a prerequisite for deployment.
Jeetu PatelPresident and Chief Product Officer · Executive, Cisco
Social post: Jeetu Patel's LinkedIn post about modernizing Cisco's code, LinkedIn, November 2, 2026.
Most people misunderstand this as tokenmaxxing. It is actually evalsmaxxing. We’re in the early stages here and I love that at Cisco we’re on the frontier and learning how to best shape the future of software engineering.
Jeetu PatelPresident and Chief Product Officer · Executive, Cisco
Social post: Jeetu Patel's LinkedIn post about modernizing Cisco's code, LinkedIn, November 2, 2026.
Something wrong? Report a correction. JSON: /api/v1/passages/cisco-running-experiment-deploying-its-output-separate-2026.json