The New Learning Model For AI-First Work
Reading Time: 12 Minutes
Image created by Superhumxn team.
Overhaul
How we learn is broken
In early 2025, Klarna’s CEO, Sebastian Siemiatkowski placed a hiring freeze on the firm. At the same time, he announced that all non-technical employees must build automated text and support frameworks in a matter of weeks.
Work is changing faster than we can process, and this is just the beginning. Until recent years, the idea of following a career ladder was generally accepted. Career growth followed a pretty straightforward and predictable path. You learned a skill, specialised over time, and gradually progressed through a series of defined roles. That model worked reasonably well when industries changed steadily and expertise remained valuable for long periods of time.
Fast-forward to 2025, and AI is rewriting the rules on what skills matter, which jobs last, and how organisations are run. And the pace of change is not slowing down any time soon. Which raises the question: how do companies keep adapting and learning at lightspeed, without breaking?
The answer isn't to learn more. I’m sure you agree that most professionals are already learning constantly. The problem runs deeper than that.
Many workplace learning systems were designed around stability rather than constant change. Traditional career development models still train employees under the premise that knowledge is fairly fixed, expertise compounds gradually, and learning is a distinct phase that happens separately from work itself.
What can I expect from a subscription?
Real-world strategies for navigating the future of work.
AI AGENTS AT WORK
Everything you need to know about how AI agents are changing teams and ways of working.
SUPERHUMXN IRL
Real insights from investors and operators.
MEMOS ON PEOPLE, CULTURE & AI AT WORK
The latest use cases and strategies for change and transformation.
Acceptance
Uncertainty is certain
Like Klarna, the fastest-moving companies today have already accepted this uncomfortable truth. Learning isn’t something that happens alongside work. It's built into how work happens. Adaptation isn't an add-on. It's the job.
That shift has consequences beyond individual habits. It changes how companies think about onboarding, hiring, leadership development, internal mobility, and performance. You can't run an organisation modelled for a slower world. Employees are expected to learn while delivering, adapt while executing, and build capability without stepping away from the work for long stretches.
So we started digging. What we found kept pointing to the same problem.
Most workplace learning systems can't keep up. Employees finish training and forget it. And when workloads mount, it's the first thing that falls away.
That's what led us to the Development LOOP model.
The premise is simple: learning works better when people can apply, repeat, and reflect on it in real time, building momentum through small behavioural adjustments rather than large bursts of information. Instead of separating development from execution, the model treats learning as something that happens inside the work itself.
Many use the terms “AI-first” and “Ai-native” as interchangeable. Although similar, Harvard Business School Online describes the difference between the concepts.
Skill Liquidity
Why Adaptability Replaces Static Expertise
Most companies are so caught up in day-to-day firefighting that they never step back to fix the system causing the fires. Companies like Netflix, Spotify, and Atlassian spotted this early. They ditched rigid career tracks and invested in mobility, cross-functional exposure, and fast feedback instead. Growth stopped being a slow climb through fixed routes.
MIT Sloan's Workforce Ecosystems research shows why this matters at a structural level. The companies managing talent most effectively aren't thinking in terms of fixed roles and internal headcount anymore. They're coordinating across employees, contractors, and external contributors as a single fluid system.
What that requires, at an individual level, is what I'd call skill liquidity, or the ability to navigate tools, contexts, and disciplines without needing a complete reset every time. Staying in one lane is increasingly a liability. The people thriving right now aren't necessarily the most experienced. They're the most liquid.
Depth/Value
Why adaptability is the game
For a long time, depth was the game. The more specialised you were, the more valuable you became.
For now, that's still true in some fields. But in most, the boundaries have blurred. Someone in marketing now needs automation knowledge, data literacy, and AI skills alongside the creative stuff. Product, strategy, design, and ops increasingly overlap. Job roles are less defined than they used to be, and that's not a bug; it's literally how the most successful startups are designed.
As a result, many companies are placing greater value on learning capacity, experimentation, interdisciplinary collaboration, and the ability to adjust quickly. Employees who can absorb ambiguity, test ideas rapidly, and refine their thinking continuously are more valuable than static experts. That does not mean deep knowledge is irrelevant. It means expertise alone is no longer enough.
Kolb’s Experiential Cycle was developed during a time of stable and predictable growth. While David Kolb’s experiential learning cycle remains foundational, the pace and unpredictability of the age of AI expose the limits of linear learning models, creating a need for more adaptive approaches built around continuous feedback, experimentation, and particularly unlearning.
LOOP
A different approach to learning
Information has never been more accessible. The hard part about learning isn't finding knowledge anymore. It's knowing what to do with it. What to learn deeply, what to skim, what to turn into a system, and what still requires real human thinking. And then being able to unlearn it just as fast when something better comes along.
That's a different skill from anything traditional learning models were built for, and it's what the Development LOOP was modelled around. Instead of treating learning as an isolated activity, the LOOP treats it as an ongoing cycle embedded directly into work.
The LOOP model is built around four stages:
L — Lay Slow down before you act. Clarify what's actually happening, question your assumptions, and identify where outdated thinking might already be shaping your decisions.
O — Operate Test ideas in the real world. Stop theorising and start moving. Experimentation and practical application create feedback that theory alone never will.
O — Organise Make sense of what you're learning. Identify patterns, extract insight, and understand what's actually working before jumping into the next experiment.
P — Propel Apply what you've learned consistently. Refine your systems, improve your decision-making, and carry those adjustments forward into the next cycle.
How the LOOP works IRL
1. Lay
Most people jump to solutions before they fully understand the problem. And the pressure to act quickly usually makes this worse.
Take a manager struggling with communication. The instinct is to assume the problem is accountability. But dig a little deeper and it usually looks different. Decision ownership is unclear. Feedback is inconsistent. People aren't raising problems because psychological safety is weak.
The real problem was never accountability. It was the system around it.
That's what the Lay stage is for. Not to slow everything down, but to make sure you're solving the right thing before you start moving.
2. Operate
This is where you stop consuming and start moving. The goal isn't perfection. It's creating enough movement for useful feedback to emerge. Test a new approach to a sales conversation. Restructure how a key decision gets made. Pilot a different way of onboarding someone. Pick something small and run it.
Most growth only becomes visible once action begins. You rarely think your way into new behaviour. You act your way into it.
3. Organise
Most teams move from one experiment straight into the next without stopping to make sense of what just happened. Work keeps moving. Understanding doesn't always follow.
That's where reflection comes in. Not as a passive exercise, but as a practical way of spotting patterns, extracting lessons, and updating assumptions based on what actually happened rather than what you expected.
Without it, experience becomes noise. You repeat mistakes, chase short-term signals, and confuse temporary momentum for real progress.
Speed doesn't automatically produce learning. Reflection on experience does.
4. Propel
This is where learning compounds. Once you know what's working, systemise it. Automate the repetitive decisions. Document the processes that keep getting reinvented. Build structure around the behaviours that consistently produce results.
Every cycle should create stronger systems, which creates less friction in the next cycle, which creates more space to adapt, experiment, and grow.
Act
Learning by doing still matters most
There's a reason internships exist. And apprenticeships. And on-the-job training. Humans have always learned better by doing than by sitting in a classroom.
David Kolb's experiential learning theory formalised that idea. People learn more effectively through active application than passive consumption. Learning lasts when action is connected to reflection on experience. The LOOP draws on that same thinking, applied to the reality of work in the age of AI.
Micro-learning
Why modern teams need shorter learning cycles
Most people are already at capacity with notifications, meetings, decisions, and context-switching. Adding a large learning programme on top of that only fuels burnout, not adaptability.
Shorter learning bursts work better because they create less resistance. When progress feels achievable, change feels manageable. And because the bar to start is lower, people actually do.
That's why micro-learning has taken off. Instead of pulling people out of their work for hours-long training sessions, learning happens in shorter, more frequent moments embedded directly into the day. Deloitte found that employees have just 4.8 minutes per day available for learning and development.
It’s no surprise that micro-learning completion rates sit at 83%, compared to 20-30% for conventional programmes. Small wins really do add up.
Culture
How adaptive learning cultures work IRL
Microsoft is probably the clearest large-scale example of an adaptive learning culture.
Before Satya Nadella took over in 2014, the company had a well-documented internal problem. A stack ranking system forced managers to rate employees against each other. Engineers avoided working with talented colleagues to protect their own rankings. Ideas died because visible failure was professionally dangerous.
Nadella scrapped it. He went on a listening tour, asked questions publicly, and acknowledged his own mistakes rather than projecting certainty. Forced ranking was gone within his first year. Performance conversations shifted toward impact, growth, and mobility. The phrase that stuck was simple: stop being a "know-it-all." Start being a "learn-it-all."
Microsoft went from a $300 billion company considered past its prime to a $3 trillion industry leader. The culture change came before the financial one.
Change
How AI changes learning
Most of the tasks that used to take time and expertise can now be handled faster by automation.
The skill that holds value now is the discernment of knowing which tasks to hand off to AI and which ones still need a human making the call. That judgement cannot be downloaded or prompted, and it develops through experience.
Habit Formation Cycle. Habit loops explain why small, repeated actions stick faster than big behavioural overhauls. This is the same principle that makes embedded learning more effective than one-off training.
Final thoughts
The idea of the career ladder hasn't completely disappeared yet. But it has stopped being a reliable map for career growth in the age of AI.
LinkedIn's 2025 Workplace Learning Report found that 49% of L&D professionals now perceive a skills crisis. Not a knowledge shortage. A skills crisis. The information exists, but the ability to know what to do with it, what to delegate to a system and what still requires human judgement, is what's in short supply.
The Development LOOP highlights this gap, encouraging a more streamlined model of learning that builds the cognitive muscle for discernment. In 2024, the proportion of companies experimenting with AI went from 5% to 95% in a single year. The organisations that will pull ahead are the ones that know which tools to use, when to use them, and when to bypass them entirely.
