Copilot model
person concept tool org talk claim — click a node to jump to its page; hover an arrow for the relation
The copilot model is a paradigm for AI adoption in which AI tools are layered onto existing workflows to make individual workers incrementally more productive — commonly cited through benchmarks such as co-pilots making engineers 20% more productive. ↗
Criticism
Tom Blomfield directly challenges this framing, arguing that thinking about AI primarily as a productivity multiplier for individual contributors is "a broken way of thinking about AI." ↗ His critique is not that the productivity gains are fictitious, but that the copilot framing sets the wrong strategic ambition — it positions AI as an assistant to humans rather than as a substrate for fundamentally restructuring how a company operates.
The alternative he contrasts with the copilot model is the concept of the self-improving company: an organization that uses AI to redesign its own processes, feedback loops, and institutional knowledge rather than simply accelerating the existing work of individual employees. ↗
Points of disagreement
The copilot model remains widely cited as evidence of AI's value — the 20% engineer productivity figure is a common benchmark used to justify AI adoption. Blomfield does not dispute that such gains exist, but contends that anchoring strategy to incremental individual productivity improvements causes organizations to miss the more transformative opportunity of systemic, company-level self-improvement. ↗