The Lego Problem: Why Delegating to AI is Not the Same as Delegating to People
Molly Graham on the shifting rules of leadership and the new burnout in tech
For years, the gold standard for scaling a startup was a simple piece of advice: 'Give away your Legos.' The idea was that as a leader grows, they must relinquish control over specific tasks—the small, modular blocks of responsibility—to allow others to build. It was a lesson in trust and organizational velocity. But in 2026, the Legos are changing. We are no longer just handing blocks to junior employees; we are handing them to models that can process, execute, and iterate at speeds no human can match. This shift is creating a strange, hollow feeling in the heart of the tech industry. It is not just about job displacement, which is a matter of economics; it is about the erosion of the shared struggle that builds professional identity.
The Ghost in the Machine
When you delegate a task to a human, you are engaging in a social contract. There is a feedback loop of mentorship, error, and eventual mastery. When you delegate to an AI, that loop vanishes. The AI does not learn from your management style, nor does it feel the weight of a mistake. This creates a vacuum. Managers are finding themselves in a position where they are overseeing systems rather than people, leading to a specific brand of loneliness. The burnout sweeping through the sector isn't just from overwork; it is from the sense that the human element—the messy, unpredictable, and rewarding part of building things—is being stripped away in favour of pure, frictionless output.
Delegating to AI is fundamentally different from delegating to a human because the AI lacks the capacity for the shared struggle that builds trust.
The fear of displacement is often framed as a fight for wages, but the real loss is the loss of the 'apprenticeship' phase of a career. If the entry-level tasks—the very Legos that allow a junior engineer or marketer to learn the ropes—are handled by a model for £0.02, how do we build the next generation of experts? We risk creating a workforce of high-level architects who have never actually laid a single brick. This isn't just a training problem; it's a structural threat to the continuity of expertise.
- Identify 'Human-Only' Legos: Tasks requiring empathy, ethical judgment, or complex social negotiation.
- Protect the Apprenticeship: Ensure juniors still have manual tasks to build foundational intuition.
- Manage the Loop, Not the Task: Focus on the integration of AI output into human strategy rather than just the output itself.
The best managers right now are not those who use AI to replace their teams, but those who use it to expand the scope of what their teams can dream. They are moving away from being task-masters and toward being curators of intent. The goal is to use the efficiency of the machine to buy back the time required for the deep, human work that actually moves the needle. If we fail to make this distinction, we will find ourselves in a world of perfect efficiency and zero meaning.
Efficiency is a tool, but shared struggle is the foundation of professional growth and human connection.