AI was the talk of Enterprise Connect 2024 last week, as we knew it would be, but I’m happy to say that I came away from the event with some practical answers to go alongside the questions prompted by the big announcements and pronouncements from the keynote stage. Here are two examples:
Many enterprise IT shops are trying to answer this question as they trial the $30/month/user Microsoft digital assistant. The precise answer varies with the enterprise, of course, but I heard a common refrain start to emerge about how large enterprises are thinking about rolling out Copilot.
There seems to be an emerging best practice that requires users to take some form of training before they get a permanent Copilot license. Which makes sense because Copilot is neither cheap nor necessarily certain to deliver value to a user who has no previous experience with the technology. Past generations of UC technology may have been simple to get basic value from—turn it on, it works, play around with add-on features (and mostly quit using those features after the novelty wears off).
As far as who’s eligible, I talked to several enterprise folks who are looking at personas and roles to target initial rollouts. In our main-stage session with strategic enterprise leaders, Carlos Quintero of AdventHealth said he expects 30% - 40% of his end users will get Copilot, though of course the process is ongoing. I also spoke with a few enterprise execs who said they’ll be monitoring Copilot usage closely, and will shift licenses away from those who wind up not using the feature often or at all.
Can AI Make a Difference in the Contact Center?
Yes. This became very clear in the presentation from Michael Altieri of Medtronic. Altieri shared Medtronic’s four-year plan for improving contact center metrics, largely through the use of intelligent virtual agents (IVAs) for both voice and text. Not only did Medtronic use conversational AI to improve IVA performance, they also used generative AI, integrating with Azure AI Services to a private large language model (LLM) and the OpenAI API. This allowed for improvements including:
The results have been impressive already, just two years into the plan. Among other metrics, per-contact costs fell 19% in the first year, to $21.02, and last year dropped to $20.68; customer satisfaction increased from 4.1 in 2022 to 4.4 in 2023; misrouted calls fell from 9% to 4%; and average wait time fell to 1 minute 49 seconds, a 37% decrease from 2021. And Altieri sees the metrics continuing to improve in 2024.
Altieri offered balanced advice as takeaways. He stressed transparency in the development process and noted the importance of involving the security and privacy teams from the beginning. But he also encouraged his peers not to be timid: “Don’t be afraid to be bold in your approach to solving the challenges. Blast through the blockers.”
So those are just a couple of AI-focused highlights from Enterprise Connect 2024. The event helped me (and I think many other attendees) start to get a better handle on what enterprises need to do in order to deploy AI and—more importantly—get value from it.
One final piece of AI-related big news from the show: Enterprise Connect is launching an event this Fall devoted exclusively to AI in the enterprise. It’s called Enterprise Connect AI, and it takes place Oct. 1 – 2 at the Santa Clara Convention Center in Santa Clara, CA. The website will be up soon, and in the meantime, you can get on the mailing list for updates by clicking here.
Hope to see you in California this fall!
