Stop Funding AI Failure
Why Your AI Training Budget is a 20th-Century Sinkhole
The $1,500 Invisible Training Stipend
I recently looked back the $1,500 annual upskilling stipend I provided to my teams. On paper, it was a forward-thinking perk. But in reality, it was a failure.
Most of my team couldn’t (and didn’t) spend it. Not because they didn’t want to learn, but because the day-to-day reality of work wouldn’t let them. To use that money, they had to go offsite to a conference or sit through a three-day workshop. But the relentless pressure of meetings, deadlines, slack-fire-drills made leaving the office for three days feel like a dereliction of duty.
I’m feeling the struggle of setting daily work aside as I try to learn Claude Code. I’ll watch ten minutes of a tutorial, get a spark of insight, and then... it’s time to make dinner. Or a meeting starts. My learning is never complete because it is being squeezed into the margins of a work life built for a pre-AI tempo.
This is the friction almost every worker I know is feeling. We are getting access to AI tools and training while keeping the same 40-hour-a-week production mandates. We are telling people to innovate in the 15 minutes between Zoom calls.
Why Traditional Training Won’t Fix It
Traditional Learning & Development (L&D) is built on the concept of training events. We treat learning like a software update where you take humans offline, install the new knowledge, and reboot.
But AI training can’t operate this way. The technology is moving too fast. The pace of technology deployment is outstripping the investment in human skills.
The trap is believing that a one-time workshop or a library of AI videos will rewire years of established processes.
If you are still budgeting for learning as a moment in time, or trying to keep a library up-to-date, you are effectively funding AI failure.
Capacity Management. So Boring, Yet So Necessary.
To lead humans through AI change without chaos, we have to start talking about upskilling solutions alongside Capacity Management. This is a discipline with deep roots in Agile, but it has been largely a practice ignored by most of the organization.
The reality is that there will always be more to do than can be realistically done, and the magic of high-performing teams is the ability to raise the most important work to the top of the stack. They allocate their time judiciously. When we ignore the discipline of rigorous capacity management, allocating our time willy-nilly, learning goes to the wayside.
And we get stuck in our current ways of working.
In a Hyperadaptive organizations, learning isn’t an extra activity, but integrated into workflows. Moving to a culture where capacity is managed and learning integrated into the fabric of operations requires three fundamental shifts in how we lead:
1. Explicitly Fund the Learning Hour
If you don’t explicitly allocate hours for learning in your capacity planning, you are defaulting to failure. Leaders up-and-down the hierarchy must understand that there are real trade-offs to be made with important initiatives.
To prioritize AI literacy, some business-as-usual projects must be put on the back burner.
We must shift from event-based learning to an always-on approach, think 45 minutes, three times a week, baked into the normal job (“time for our AI Learning Power Hour!”). As Ryan Martens notes in the foreword to my upcoming book Hyperadaptive, you can’t strategize your way through a hurricane, and sometimes you have to ease off the course by twenty degrees to make actual headway.
2. Respect the J-Curve of ROI
Every leader wants immediate productivity gains from AI. Research shows that AI-augmented consultants can finish tasks 12.2% faster with 40% higher quality. But there is a J-Curve involved.
Before you get the spike in productivity, you will see a dip. People will be slower while they learn. If you measure ROI too early, you will kill the initiative. Instead, we must measure interim metrics, such as:
AI Literacy Rates: How many people can effectively use agentic systems?
Use Case Velocity: What is the quality and volume of AI use cases being surfaced from the frontline?
Cognitive Load: Are we freeing human capacity for creativity, or just piling more on?
3. Build Psychological Safety into the Calendar
One of the most effective practices I’ve seen is Failure Friday, where teams share their AI failures (some fabulous, some frustrating). If your team is popcorning AI ideas but afraid to admit when an automation fails, you won’t build the institutional capabilities around experimentations needed to survive.
We need to create practice arenas, olaces like PwC’s Prompting Parties, where employees can experiment on real work items without the fear of a botched production result. This is how you move from knowing AI to being capable with it.
Creating AI Learning Flywheels
Because of the speed of changes with AI, we need to rethink the role of L&D in the enterprise. I posit that the future of L&D looks like learning professionals embedded into a self-sustaining system. They become the AI sense-makers, monitoring AI developments, atomizing the learning, and delivering updates to the right people and the right time. In the Hyperadaptive model, we use a four-step flywheel:
Spark: Use hands-on workshops to spark AI usage and new ways to solve real business.
Spread: Identify your natural AI Leads, the enthusiasts who will teach their peers, then provide them with systematic, ongoing support.
Scale: Establish an AI Activation Hub to house best practices, atomize the learning, measure success, and keep up with the pace of change.
Sustain: Embed these loops into the organization’s DNA so learning happens faster than the world changes.
I detail this flywheel in my AI Learning Ebook
Something to Consider
We often blame culture for slow AI adoption, but culture is just a reflection of where we spend our time and money.
My question for you this week: If I looked at your team’s calendar for next Tuesday, would I see AI Learning as a protected, funded line item, or is it a ghost that only appears if they finish their real work early? What is your reality? 🤔
Explore How to Integrate AI into Your Organization
The articles in my Substack are rooted in my upcoming book, Hyperadaptive: Rewiring the Enterprise to Become AI-Native. The book comes out in May 2026, but you can get early access to the model and frameworks by pre-ordering the book.
Order via amazon, then redeem your bonus materials at: hyperadaptive.solutions/book