Prognostication vs. reality
I spent 2025 as Founder in Residence at Operator Collective, a venture firm whose LPs include many of the most successful tech leaders of the last decade. During that year, I spent time not only with hundreds of AI startups but also with these operators. For me, these discussions informed several predictions about AI that are now coming true,
So I was extra interested in the data set that Operator Collective dropped last week. They surveyed the operators in tech who are shaping the future of AI at work, from Microsoft and Google and Anthropic and OpenAI and many more, to find out what's really happening with AI transformation on the ground inside these companies.
What I thought I knew vs. what I actually know
I have spent my entire tech career in enterprise AI, first as a product manager at Microsoft, and then as the founder and CEO of Textio, which created the first AI solutions for HR teams more than a decade ago. In the last couple of years, I have published several new data sets about AI at work in the nerd processor newsletter. A bunch of my AI predictions are playing out in 2026.
Given my experience, I started the year confident about how the AI revolution in enterprise would continue to develop. But I recently jumped back into operating as the VP of AI Transformation at Microsoft, and I’ve discovered that much of what I believed was wrong.
There are three principles that I now believe are driving the future, all of which are borne out in the research published by Operator Collective last week.
#1: We are in an anarchist’s moment. Organizations are defining systems and principles for how AI should be used in the workplace. But for every thoughtful policy about privacy and security, there are 1,000 employees pasting sensitive company information into their AI chatbot of choice. The tools are changing rapidly, and they are so widely available that curious people can experiment freely.
In this context, I was not surprised to read that so many of the survey participants feel individually responsible for AI transformation. Regardless of which executive is ostensibly responsible, it is workers at all levels who are defining the future.
#2: In AI transformation, enablement is a better framework than compliance. Organizations are right to be concerned with data provenance and security. But by the time you hash out, document, and mandate your AI policies, your team has tried three more free tools. The Operator Collective results show just how many AI solutions the average organization has already tested out. It's a ton.
The best enterprise policies look out for data security, but make it easy for people to try new tools in sandboxed ways and choose the right solutions. The last thing you want is over-restrictive policies that make people afraid to try new technologies, or your innovation will fall behind.
#3: Velocity rules. Organizations have tried several metrics to measure the impact of AI use at work, ranging from sheer adoption to time and cost savings. Most of these are measurement theater and don't do much to show sustainable AI impact.
Across the board, the metrics that are rising to the top among expert operators are focused on speed, delivery, and execution. Can you ship products faster or be more responsive with customers? In other words, the right metrics to assess the impact of AI are the same metrics that boards cared about in the first place.
The bottom line: In contrast to a lot of future of work commentary, the voices in this particular survey come from operators who are leading AI transformation on the ground. The data sample is skewed to currently active tech execs, so it doesn't purport to speak to what is happening with all jobs at all career stages in all industries. To my mind, that makes it more credible, not less.
I'd love to see similar work done with people in other industries and at other career stages. Most AI hot takes skip that depth. What are doctors seeing? How about law firms and banks? To know where a particular industry is going, we have to hear from the people who work within it.
Kieran
If you liked this story, why not subscribe to nerd processor and get the back issues? Also, why not learn to tell data stories of your own?
My latest data stories | Build like a founder | nerdprocessor.com