Cisco Shares a Three-Pronged AI Strategy at WebexOne

This article was originally published on No Jitter.

At the WebexOne event in Anaheim, Calif., Cisco shared an AI strategy centered on the three pillars it has discussed for a while reimagining work, building the best collaboration suite, and building the best device portfolio.

With hybrid work, meetings will involve a mix of in-person and remote participants. Businesses are transitioning from a completely remote workforce to some kind of hybrid environment, although most have yet to figure out what the end state looks like. Cisco’s research found that 98% of all meetings will have at least one remote person moving forward. This is similar to my research showing that 75% of knowledge workers will be remote at least once a week. This creates challenges in meeting equity, which can limit participation, the ability to read body language, and more. Cisco is looking to use AI to close the gap between in-person and remote meetings.

Before WebexOne, Cisco pre-briefed the analyst community to give us more details on its strategy. Javed Khan, SVP and GM of Collaboration, stated that the company’s approach revolves around three challenges. “The three problems we are solving are still the same,” he told me. “We’re making sure that we make hybrid work, work. Second, as offices evolve, we’re making the office a magnet, not a mandate. Thirdly, enabling our customers to show to their customers that they care.”

Related:Cisco to Acquire Splunk in $28B Deal in Bid to Boost Observability, Security

While there were lots of announcements, I found the AI ones the most compelling and worth a deeper dive. This includes real-time media models (RMMs), audio codecs, Webex AI assistant, and a partnership with NVIDIA.

Real-Time Media Models Move Beyond Text Insights

“Large language models are really about things like email messages, transcripts, documents, and then training on large volumes of data,” Khan stated. “That’s what large language models do. And then we extract insights from it.”

But when it comes to collaboration, a lot more goes into a conversation than just text—there are also also reactions, gestures, tone of voice, and inflections.

RMMs in Webex will be able to produce multiple outputs, including people and object recognition, as well as analyze movement and gestures and use those as “signals of context” to enhance the common text-based capabilities we all know, such as meeting summaries and highlights. This will provide rich real-time insights. For example, Webex might see that you’ve stepped away for a moment and automatically provide meeting notes that can fill in the gaps when you return to your desk.

Some users I have talked to have complained that the collaboration vendors have introduced so many new features it’s hard to keep up. What I like about this approach is that it makes AI seamless. In this case, Webex will prompt the user without any action on the worker's part, making the experience not only easy, but effortless.

New AI Audio Codec Smooths Choppy Audio

The industry has relied on Opus, the lossy audio codec the Internet Engineering Task Force (IETF) developed for a decade. Cisco thought it was time to rethink that codec with an eye on audio and video quality. “We have spent the last two years quietly working on a next-generation AI codec,” Khan explained. “This is going to power a lot of those real-time media models.”

Choppy audio or video can create havoc in a conversation—frustrating meeting participants and customers talking to a contact center. Cisco’s AI Audio Codec in Webex is a new generative AI solution to improve the audio quality of real-time communication. The company aims to deliver “crystal-clear audio” despite poor network conditions and inconsistent connections. The company says that the efficiency of the AI Audio Codec provides considerable redundancy in transmission, which should reduce network packet loss. The new codec also has speech enhancement functions that can handle noise removal, echo, and bandwidth extension.

Cisco is using machine learning to deliver high-definition meetings regardless of bandwidth conditions. In fact, during his keynote, Cisco EVP Jeetu Patel stated the AI codec can correct network conditions with up to 90% packet loss. The ability to deliver high-quality audio and video over a poor-performing network is unique to Cisco in the communications space. The company can do this because of its network knowledge, and is a great example of how Webex on Cisco gives the vendor a broader platform advantage.

Webex AI Assistant Generates Answers in Real Time

The company also launched a set of capabilities to help minimize work and costs for IT. The Webex AI Assistant will enable users to type queries about a meeting they might have missed. For example, you might type “catch me up on the 15 minutes I missed in the meeting” or “catch me up on all the meetings I missed on Friday while I was on PTO.” Cisco says AI Assistant will generate the answers in real-time. The company says that Webex AI Assistant is already included with Cisco Collaboration devices and will be part of the entire Webex portfolio, with the ability to change message tone, use prompts for suggested message responses, meeting summaries to help users catch up, summaries of unread messages Summaries, and Slido topic summaries.

It's important to note that because this is done with generative AI, natural language becomes the user interface, dramatically lowering the barrier to entry.

Cisco says these new capabilities should all ship before the end of 2023.

Cisco Announces Strategic Partnership with NVIDIA

At the event, Cisco announced a partnership with NVIDIA. I was surprised by the announcement, as Cisco has been using NVIDIA GPUs for about a decade. At the event, I asked SVP and GM of Devices Snorre Kjesbu about this, and he told me although Cisco has been using NVIDIA chips for a long time, they never announced a formal partnership, and the company felt with AI becoming a bigger part of device design, it made sense to do so now.

He also explained that traditionally, Cisco rarely exposes the components that are “under the hood” as they want the Cisco brand to be front and center. Over the years the relationship has changed from being silicon only to silicon, software, and a valued discussion partner to help with innovation. Kjesbu added, “NVIDIA is incredible to work with. They are agile, nimble and have a great understanding of technology trends and how accelerated computing can solve the difficult problems. Given the maturation of the partnership, it seemed time to formally announce it.”

As part of the NVIDIA announcements, Cisco took the covers off the new Room KIT EQX, the most recent device using the NVIDIA Jetson Xavier Edge AI platform. The NVIDIA silicon powers many advanced, on-device video, audio, and speech intelligence capabilities. During his keynote, Kjesbu introduced the concept of “Distance Zero” where users will eventually feel like they are in the same room as someone regardless of how far they are and the devices core to that vision.

The industry is being hammered with AI announcements, and, as one would expect, some are real, but others are vaporware. Cisco’s AI strategy is multi-pronged and can deliver value today but exponential value as Cisco ties all its touch points together. It's fair to say that Webex was behind the curve on the cloud, but it’s looking to use AI to move ahead of the curve. The AI race is on, and Cisco’s massive size and breadth of portfolio should give it an advantage.