Agents of Innovation/Learning/Destruction/Creation
The difference between a challenge and an opportunity is often perspective
I wrote about using AI to build “digital twins” of literally everyone in your company a while back. I also wrote about how I thought #LearningAndDevelopment orgs had a chance to move away from thinking about the output of their efforts (activity = courses) to what orgs are really after (value = answers). Then I read this post by the brilliant Alan Lepofsky, in which he writes (about search) “These search engines (or answer assistants), respond when we prompt them, but they are not yet proactively providing us information in the context or flow of the moment.” Add to all that the truly groundbreaking work that Mike Hruska is doing with agents and augmented organizations and it all makes me feel hopeful…but if I owned a company in the learning space, I might be a skoosh worried.
Little story (You know how I love em) - when I was interviewing to come to work at Amazon, I also interviewed at two other places that trip - Sears (we can talk about that later) and Twitter. I remember coming out of Twitter and thinking it was cool (I was a heavy user at the time) and they had that SF thing going on, and the food was awesome but I also remember thinking that two people with an AWS account could build this (ignoring the need to build the network effects). Amazon felt it had a much more defensible moat. I think along those same lines, a lot of the companies in the learning space are having their moats greatly reduced in their ability to defend against newcomers and here’s what I mean.
We start with agents - no, not like Jerry McGuire but AI Agents. The simplest way to think about an AI agent is that it represents an automated function that can operate with some degree of autonomy. Now we’ve also thought of agents as bots but given advances in areas like Natural Language Processing and foundational areas of machine learning, are giving agents the ability to leap forward in terms of what they can do for us.
Let’s think about my ‘digital twin for everyone’ idea. As that twin gains greater resolution - as you interact more with systems and data, your “Learning Agent” is also gaining greater insight and resolution on not just the things you will need to know but exactly when you will need to know them. It will also know your preferred channel for receiving that information or put a different way, it will know the best place in your personal workflow, to inject that knowledge. Your agent, because it will be part of the agent collective intelligence that extends throughout the organization, will also be able to tell you who else is working on the same issues, the solutions, they found and if they’re actually sitting in the office next to you.
Now let’s come back to what traditional L&D teams regard as their value. The nomenclature is usually something around “performance” and the struggle is always to link that training content to increases in performance. I’d argue that what is really meant are courses. Designers. Developers. SMEs. Administrators. and more. there’s a whole Great Chain of Being built to produce courses. Agents could also produce courses. If you go to There’s An AI For That and search for AIs that are being deployed to build or help build online courses, you’re going to be able to scroll through multiple pages of tools for that task. Now remember Ethan Mollick’s admonition that the AI we use today will be the worst AI we will ever use and understand that none of those AIs existed more than 2 years ago and you can see how fast this line is advancing.
So agents can build courses but more importantly, to Alan’s point at the start of this essay, is that agents can provide answers. How many times, when confronted with let’s say a time sensitive problem, have you heard people utter the words “what I really need is a course”? Agents will have the ability to know when you’re in a time sensitive situation, they’ll be trained on the same corpus of corporate data that you are so they’ll have access to that - they’ll know your calendar and see your IMs and emails and will be able to do sentiment analysis in real time and understand urgency and eventually they’ll be able to intuit what you need, how you want it, and when you reach out your hand, they will place the answer in your palm just when you need it.
If you say that they can’t do this today - I’d argue that you’re only half right but even more that if you wait 12 months, that argument will be moot. The point to all of this is that we can look at it as existential threat or an amazing opportunity. If its the latter, then we need to be thinking about what our true value is versus our activity and how we will contribute that value, literally what mechanisms will we use, when agents reach that point. We have a chance to move up that Great Chain of Being (Value) but we need to start thinking now about what that looks like.