November 2, 2023
By Bob O'Donnell
There’s no question that Generative AI (GenAI) applications are starting to make a big impact on enterprise applications. Until recently, however, the vast majority of the attention being given to GenAI applications for business has had to do with text-based Large Language Models (LLMs) and chatbot-style applications that leverage them. At their recent Webex One conference, Cisco made a point to emphasize how foundation models and other GenAI tools can make a powerful impact in communications and collaboration as well.
The company made several announcements regarding AI integration into its entire suite of Webex tools at the event. Most involve an additional level of intelligence being integrated into Webex, but some cover audio and video quality improvements. In addition, Cisco added to its history of Apple collaborations by unveiling additional support for Apple devices, including Apple TV and Apple Watch.
Of course, like all its competitors, Cisco first started offering AI-powered features in Webex, such as automatic transcription and meeting summarization, a while back. The capabilities they just announced, however, show a new level of sophistication and provide an interesting pointer towards how GenAI-based tools are likely to evolve.
Undoubtedly the most intriguing is the introduction of a new type of foundation model Cisco is calling Real-Time Media Models (RMMs) that are based on audio and video data. According to Cisco, these RMMs can provide supplemental data and information about a meeting from things like visual cues, gestures, body language, message tone and more. As useful as text-based meeting transcriptions and summaries can be, they don’t always capture the subtle but important nuances in how things are said and how people react, both of which can have a huge influence on how certain ideas or messages are understood and conveyed. It’s this kind of information that Cisco believes it can bring to its existing AI-powered meeting notes by combining the RMM data with its existing LLM-based text contact.
In theory, these kinds of GenAI-enabled capabilities sound quite impressive, but it will take some real-world usage to see how effective they are in capturing and integrating this information into meeting summaries. Regardless, it’s an interesting example of how Webex continues to innovate in unique ways that help it stay a step ahead of its competition—not an easy task in a market that moves as quickly as video collaboration.
One of the other benefits of leveraging RMMs, according to Cisco, is the ability to create meeting summarization tools in video form. This allows people to get a movie preview style summary of a meeting, allowing people to easily watch what happened instead of simply reading about it. Again, the proof will be in the pudding, but it’s unquestionably a great concept.
RMMs also play a role in the Webex AI Assistant. Like other digital assistant-type features, the Webex version is designed to provide intelligent responses to queries that users can type into chat-style interface. For example, people can ask the Webex AI Assistant to summarize the results of all the meetings they have missed on a PTO day or highlight the comments their boss made during a particular meeting and so on. The difference versus similar types of features the company has introduced in the past is that the integration of the RMM data should provide the kind of additional context referred to earlier. Leveraging the RMM video data, future versions of the Webex AI Assistant might also be able to do things like automatically generate summaries of the portions of a meeting a person may have been missed because of running late or having to step out for a bit. The idea is that by analyzing the video feed, Webex will be able to determine when a person was there or wasn’t and then fill in those missing time gaps in a seamless manner.
The Webex AI Assistant also has the ability to provide suggested answers for customer service agents, helping them improve their responses, and can even make suggestions in the tone of how things are (or should be) said. One of the more intriguing aspects of Cisco’s implementation is that they said they intend to offer the ability to support several different LLMs in addition to their own RMMs in the Webex AI Assistant. In other words, they can essentially swap out the GenAI engine powering the experience in order to pick the best choice for different customers, different types of employees, etc. Again, this idea of model swapping is an intriguing concept that I expect to become more common in other types of GenAI-powered applications and services, so it’s very interesting to see Cisco jumping into this new realm so early.
The final piece of the new Webex AI strategy is an enhanced set of AI audio and video codecs that are designed to improve the audio and video quality on video calls where the bandwidth is limited. First, these new tools can reduce the amount of data needed down to 1/16th of what it currently is, making it easier to work in low bandwidth environments. In addition, by leveraging GenAI, these new codecs can do things like help recreate lost packets to help ensure a smooth audio signal. On the video side, by utilizing super resolution technologies like what’s been deployed for AI-generated image enhancement, Webex can upscale the quality and video resolution of a video signal over a low quality or low bandwidth connection. The net result is a smoother, more seamless experience that makes everyone in the meeting look and sound better and avoids the annoying glitches that can sometimes occur in these types of situations.
The evolution of meeting and collaboration tools like Webex and its competitors continues at a breakneck pace and while Cisco has once again jumped ahead in terms of innovative capabilities, it may not be long before companies like Microsoft, Zoom and Google start to catch up. Still, it’s great to see how Cisco is pushing forward the integration of GenAI into its Webex suite and providing an intriguing view into how these technologies continue to evolve.
Here's a link to the original column: https://www.linkedin.com/pulse/ciscos-webex-extends-generative-ai-collaboration-bob-o-donnell-cwbcc/ Bob O’Donnell is the president and chief analyst of TECHnalysis Research, LLC a market research firm that provides strategic consulting and market research services to the technology industry and professional financial community. You can follow him on LinkedIn at Bob O’Donnell or on Twitter @bobodtech. |
