‘I’m Trying To Replace Myself’: A Cisco AI Leader On Finding The Boundaries Of Automation

*All opinions expressed in this piece are those of Vimarsh Puneet, and do not necessarily reflect the views of his employer.

The most valuable currency in enterprise AI often is not the technology itself, but a sober understanding of what it can and cannot do. As companies race to automate, optimize, and generate, a simple truth is emerging: the real differentiators will come not from the most powerful models, but from prioritizing the right data, aligning AI with real customer needs, and preserving human context where it matters most. The narrative of replacement is giving way to a more nuanced reality where AI’s greatest promise is in reshaping what humans do best.

We spoke with Vimarsh Puneet, Director of Product Management for AI at Cisco. His perspective is forged by a unique career trajectory that saw him rise from a Senior Software Engineer building the technology to a product leader defining its purpose.

Puneet argued that the industry's obsession with technological capability often misses the fundamental point of business of solving a human problem. The conversation isn't about what AI will replace, but what it will elevate.

• A new human purpose: "AI isn't about replacing us," Puneet said. "It's reshaping what humans can do best: empathize, ideate, and build trust." This human-centric view is grounded in a pragmatic assessment of where AI excels and, more importantly, where it fails. For Puneet, the path to unlocking AI’s true value begins with a return to first principles, starting with the one asset that models alone cannot replicate. As we’ve explored previously, many companies are sitting on a goldmine of data without even realizing it.

Data is still the king no matter what. If you have the right data, you can actually produce way better results," Puneet explained. He points to the foundational problem with applying generalist models to specific business challenges.

• Data is still king: "The data was not written for the AI to read and learn. It was written for a human who had some back knowledge. A small, disclaimer note might state that a platform is supported, but it was written in a way that the AI would not read it right. A human would know, 'Okay, this is the first sentence I should read, and therefore it is the most important one.'"