I attended Microsoft Ignite in Seattle last week to see the AI and Copilot launch on Azure and Windows.
While watching the presentation, I was reminded of the launch of Microsoft Bob in the 1990s and the first attempt at creating a digital assistant called Clippy. Neither of these products met expectations. Copilot is far more powerful than these two previous offerings, and its power, both within and outside Microsoft, cannot be overstated.
Before the event, I met with Reply. This company specializes in preparing companies for Copilot and setting up metrics to confirm the benefits. Reply raved about how much more productive their company and clients were.
It is amazing to think that Copilot was only eight months old. It was made available to all the top companies in medicine, agriculture, services, software, and banking. Now, 70% of users say they won’t work without the capability.
I have been working in the technology industry for 40 years, and I’ve never witnessed a technology develop so quickly from a beta to what it is now. This week, let’s discuss that.
Fisker Ocean is my product of the week. I just ordered this electric SUV, and I believe it’s the best that you can buy before 2025. The Fisker Ocean is the Product of the Week. I just ordered it and think that this electric SUV will be one of the best you can buy before 2025.
The Legacy of Bob & Clippy
Microsoft was ahead of the game with Bob and Clippy. Microsoft correctly determined that users wanted a computing environment where the computer did most of the heavy lifting. Bob was a design with an avatar that, in theory, could take English commands and do what you asked. Clippy, an assistant on Windows that did the same without the avatar front end, had a similar design.
Bob’s sad but funny side was that it was originally designed for older people who could not or did not want to learn Windows. When surveyed, it was found that this group liked the product. Microsoft’s product department tried to position it as the next-generation Windows. But people who use Windows do not want it.
AI was not yet available. Clippy Bob and Bob were command-oriented, scripted offerings. They set the expectation of being able to chat with your computer, but they failed when they could not meet this expectation.
Conversational AI allows you to talk to your computer as though you were speaking to a human. This technology not only allows you to converse more effectively with your computer but also learns your working style and can do repetitive tasks for you, such as answering emails, scheduling meetings, and filling in during team meetings.
Steve Ballmer Problems: How to Fix Them
As I watched this presentation, I remembered a meeting I had with my former Microsoft CEO, Steve Ballmer. He struggled to prioritize customer requests. He was unable to prioritize customer requests because he had so many national and multi-national customers.
Ballmer was one of the smartest people I have ever known. He excelled at statistics, but the complexity and difficulty of the problem were beyond his abilities.
Microsoft showcased Copilot for Service at Ignite. This allows a greater amount of feedback to be collected from all customers, not just those with the most significant amounts. This resulted in a prioritized plan based on the user-defined criteria. The customer then validated this action plan.
If Ballmer had used Copilot, he would have had a much better career, and Microsoft products would have been more appealing during his tenure.
Ballmer’s issue was a problem that conventional methods of analysis could not solve because of the volume and speed with which data changes.
AI now works at machine speed. It can provide answers in a fraction of the time that conventional methods take. This means that executives with marginal performance could use AI to become stars, and CEOs who are unbeatable could use it to become superstars.
This is a game-changer.
Reply’s Copilot beta experience
In the same way, as I mentioned in my opening, I met Reply the day prior to the conference. Reply has worked with many customers to put this beta version into production. A product update caused the problem about three weeks ago. As this incident shows, deploying a beta product is always risky.
You’d expect people to say that it’s beta and a fix will be released soon. However, more than two-thirds of users said they could not live without it and were already dependent on it. They needed and wanted a fix ASAP and got it. I can’t remember ever witnessing this much anguish over a bug in beta software.
Initial feedback indicates a productivity improvement of between 10 and 30 percent.
These stats do not show that the improvement in productivity is due to Copilot automating the tedious administrative tasks that employees dislike, such as responding to emails, making notes during meetings, creating an agenda, messing around with Excel, and trying to remember the basics of pivot tables. Copilot did a lot of this for them when it was in beta while users were still learning how to use the product.
Professor Productollick of Wharton, who has conducted extensive research into generative AI, said that once users are familiarized with tools such as Copilot, productivity increases could jump up to 80%. This is unprecedented. Microsoft and Reply may have had impressive initial statistics, but by next year, those numbers could be unimaginable.
The technology will also advance at an unprecedented rate. Not only will the employees who use this technology improve massively, but I have never seen anything like it. This figure could be exceeded by the combined improvements of AI and users, as the previous estimate was based solely on improvements made by the users.