5 takeaways from the Cisco AI Summit5 takeaways from the Cisco AI Summit

In January 2025, Cisco held its inaugural AI Summit in Silicon Valley, where tech leaders discussed the opportunities and ethics around AI.

The event was purposely small to provide more value to attendees, primarily technology executives from major brands. During his opening keynotes, Cisco CEO Chuck Robbins noted that 40 of the Fortune 100 were represented at the event. The lineup was an appealing mix of speakers from Scale AI, Glean, Credo AI, Cohere, OpenAI, Goldman Sachs and more, as well as many Cisco executives.

Attending the event changed my thinking in some areas of AI and confirmed other thoughts. Below are my top five takeaways from the Cisco AI Summit.

1. AI is going to displace jobs but create new ones

The most common question people ask me about AI is how it will affect jobs. Many people believe it will replace jobs, while others believe it will augment roles. Like all tech transitions, the reality is it will displace many jobs. VoIP did away with voice engineers; PCs effectively eradicated mainframe administrators; the cloud lessened the reliance on server administrators; and AI will have a similar effect. However, each technology shift created more jobs at the back end of the transition.

In 2023, the World Economic Forum predicted that AI would eliminate 85 million jobs but create 97 million. Right now, we aren't sure what those jobs will be, which results in the nervousness around AI we see today. During the summit, David Solomon, CEO of Goldman Sachs, acknowledged this transition.

"Big technology waves have brought big changes in the world for a long period of time," Solomon said. "They destroy jobs, yet we managed to create new jobs, and we drive growth in the economy."

The key to transitioning from the pre-AI world to post-AI is embracing the technology and adapting to change.

2. We are running out of real data to train AI models, but synthetic data can provide relief

I first heard this statement about a year ago, and I wasn't sure how it was possible, given how much data is created with AI. During the summit, Scale AI CEO Alexandr Wang reiterated this sentiment and voiced his concern about AI being able to evolve at the speed we've seen.

It's essential to realize that AI models are only trained with publicly available data, and much of the data being created is proprietary. As these systems grow in complexity, the amount of high-quality, labeled data needed to train them effectively grows exponentially.

Synthetic data offers a solution by using algorithms to generate artificial data that mimic real-world scenarios, providing a virtually limitless resource for training AI models. This approach can fill gaps where real data is sparse or hard to collect, allowing for more diverse training data without compromising privacy or ethical concerns.

3. AI pricing models are still up in the air

Another common question about AI is how the pricing model works. Most software is currently sold with a subscription model where the business pays per user, per month, with cloud services typically based on utilization. How AI should be priced is a bit up in the air. CEOs and CFOs frequently ask this question on earnings calls because how a company monetizes AI can affect its valuation.

A company with a subscription model has a predictable revenue stream, whereas utilization-based pricing is more unpredictable. During a fireside chat at the Cisco AI Summit, Box CEO Aaron Levie offered another option: outcome-based pricing. He gave an example of an AI tool designed to generate leads, and a vendor could charge based on the quality of the leads generated.

There appears to be no consensus on how AI-based services will be priced, except that change is coming. IT leaders must press their vendors to ensure they fully understand how and when change is coming.

4. AI will create an incremental step function in productivity and change worker expectations

Many of the sessions discussed how people use AI. Box's Levie discussed how AI is shifting from assisting users in doing their jobs to being able to do the work on behalf of someone. In a separate discussion, Arvind Jain, CEO of Glean, discussed how the success of AI resides in its ability to complete complex end-to-end tasks.

This means it won't always replace people but allow them to do more. In a contact center, an AI agent can summarize and log call summaries faster and more accurately than people. In a sales department, an AI tool can analyze calls and emails and help qualify prospects faster. In healthcare, AI agents can diagnose MRIs faster, enabling doctors to spend more time treating patients.

What do people do with the extra time? Do they take Fridays off? Do they enjoy longer lunches? While this might sound appealing, we can likely expect higher productivity levels, and businesses should adjust their goals accordingly. In the examples above, a contact center agent should be expected to handle more calls, while the salesperson would be expected to close more deals. As Solomon stated, "We drive growth in the economy."

We've seen this pattern before. Automation tools raised sales quotas, and analytics made banking services, such as mortgage approvals, faster. The use of AI will enable employees to work faster and smarter, and productivity expectations will grow as a result.

5. AI requires a security rethink

The primary purpose of the Cisco AI Summit was to drive thought leadership in AI. Its secondary purpose was to launch AI Defense, a security tool designed to protect companies from the unique risks associated with AI.

Security has always been a challenge for companies, as the tech environment continues to evolve in ways that expand the attack surface. Cloud, mobile, work from home, SaaS and other trends have made the technology environment more dynamic and distributed, which plays havoc with cyber teams.

Cisco's AI Defense secures AI by providing visibility into AI applications used within an enterprise. It validates the various AI models for potential vulnerabilities. Organizations can use AI Defense to enforce security controls and safeguard the company from AI misuse and data leakage.

In a recent conversation I had with fellow analyst Bob O'Donnell from TECHnalysis Research, he explained, "Cisco is using observability tools, such as Splunk and ThousandEyes, to provide insights into what models are running, the types of queries being run, the prompts being used and extend the notion of observability into the AI era."

He said this move "helps position Cisco as an AI thought leader." More thought leadership from the mainstream IT vendors is good for the industry. To date, most AI evangelism has come from Nvidia and startups. Major IT vendors such as Cisco will play a key role as AI scales.

Summary

The Cisco AI Summit accomplished its primary goal of providing a thought leadership platform for AI. The company stated it would make the event an annual one, which I was glad to hear, as tech needs more industry events rather than ones overloaded with product information.