Conversation 1: On AI and Headcount

The question you will hear: "Why can't we cut engineers now that AI writes code?"

Your stance: Artificial intelligence does not eliminate the need for engineers. It transforms what engineers are for.

Before AI coding tools, the bottleneck in software development was writing code—typing speed, syntax knowledge, turning requirements into working functions. That is what organizations hired for. Now the bottleneck is judging code. When a machine can write a feature in minutes, the scarce skill becomes knowing if that feature is correct, secure, maintainable, and aligned with the architecture.

Cutting headcount in this environment does not produce the same output with fewer people. It produces a codebase that nobody understands at depth. It produces security postures that decay within months. It produces production environments where every incident takes dramatically longer to resolve because the people who would have built the systems never built them.

How to lead in that room: Do not defend the current headcount. Bring a proposal for what the team looks like under new conditions. That may mean a smaller, more senior team with different skill profiles—engineers who specialize in evaluation, integration, and systems thinking. It may mean the same headcount with a radically different composition, adding AI operations specialists, model evaluators, and platform architects.

If cuts are non-negotiable, make the trade-off visible: here is what we stop maintaining, here is the risk we accept, and here is what will break. Shift from saying no to framing choices between imperfect options.

Before	After
Engineers hired and measured by coding speed	Engineers hired and measured by judgment and systems thinking
Headcount based on feature roadmap estimates	Headcount based on risk tolerance and cognitive load
AI tools as individual productivity gains	AI tools as team-wide accelerators that increase review load
Technology leader fights to preserve every role	Technology leader proposes new team shapes aligned to new bottlenecks
Reduction as a math problem	Reduction as a risk decision with visible consequences

What to avoid: Defending current headcount as a right. Dismissing AI productivity gains as mythical. Letting the conversation become about you vs. the CFO.