Slide 01

The People Conversation

CxO + Director + Board

Thesis

Executives focus on tooling and training. The real question is harder: can your people actually do this? AI agents do not fix subpar engineering. They amplify whatever is already there.

Strong engineers using agents become dramatically more productive. Weaker engineers produce confused output faster. The difference is not the tool. The difference is the person holding it.

Kindness Being kind is not pretending the change is not happening. It is telling people the truth early enough that they can act on it.

Slide 02

AI Does Not Fix Subpar Engineering. It Amplifies Whatever Is Already There.

Core problem

Strong engineers with agents

Externalize requirements clearly because they understand what they are building and why.
Evaluate agent output against system constraints, security boundaries, and performance characteristics.
Override agent decisions when the generated code misses domain nuance or introduces subtle defects.
Ship production-grade software daily because they know what good looks like.

Weak engineers with agents

Cannot articulate what they need because they built careers on pattern-matching and copying code.
Accept agent output uncritically because they lack the foundation to evaluate it.
Produce confused output faster, shipping bugs at velocity instead of features.
Mistake volume for quality. More code is not better code. It is more surface area for failure.

Slide 03

Fewer Handoffs, Flatter Teams, Compounding Quarterly Advantage

Economics

Team composition shift 8 not 40

Eight empowered people ship faster than forty with approval gates, context-translation layers, and dependency negotiations.

Advantage cadence Quarterly

Organizations pursuing AI-native velocity compound their lead every quarter. The gap is not linear. It accelerates.

Leadership change Hands on

CTOs and directors writing production code again. Agents removed the drudgery that drove experienced leaders into management.

Fred Brooks told us in 1975 that adding people to a late project makes it later. We spent fifty years ignoring him because headcount felt like progress. Economics, not philosophy, is ending that era.

The Mythical Man-Month, reframed

Slide 04

A Three-Phase Framework to Know Who Can Make This Transition

Operating model

Phase 1

Written exam: one hour, closed-book, no AI

Data structures, performance characteristics, SOLID principles, design patterns. Foundational knowledge without external aids. Cannot pass this? Cannot evaluate what an agent generates.

Phase 2

Whiteboard design: ninety minutes, in person

Design a complete system -- payment platform, ticketing engine. Evaluate system thinking, boundary identification, complexity reasoning, tradeoff explanation. Cannot do this? Cannot direct an agent effectively.

Phase 3

Agent-assisted build: six hours, AI agent of choice

Working endpoints, data model, tests, deployable artifact. Assess code comprehension, output evaluation, override decisions. Cannot ship working software with an agent? That is the conversation you need to have.

Purpose This is not a gotcha. It is a map. People who struggle in Phase 1 need foundation work. People who struggle in Phase 2 need mentorship. People who struggle in Phase 3 need practice. All of this is addressable -- if you start now.

Slide 05

Invest in Your People Before the Market Makes the Decision for You

Implementation

What this looks like in practice

Run the framework. Invest in development. Restructure teams around capability, not headcount.

Engineers who built expertise in pre-AI systems rewarded specific behaviors: deep framework knowledge, pattern replication, process compliance. Those behaviors were rational then. They are insufficient now.

The skills that matter -- externalizing reasoning, evaluating generated output, directing agent workflows -- were never selected for. But they are learnable. Through deliberate practice, mentorship, and honest feedback.

Investment Foundation workshops for Phase 1 gaps. Design mentorship for Phase 2. Paired agent sessions for Phase 3. Twelve weeks to material improvement.

What honesty requires

Some people will not make this transition. Not because they are not talented. Because the work changed and their skills did not. Telling them early is kind. Letting them discover it through a layoff is not.

Transition support, reskilling budgets, generous timelines -- these are what a good organization provides. But pretending the change is not happening helps no one.

Careers People deserve to know where they stand while they still have time and options. That is what this framework gives them.

Slide 06

The Timer Is Not on Your Technology. The Timer Is on Your People.

Decision close

The countdown

Organizations that assess, invest, and restructure now will compound their advantage every quarter. Those that delay will discover who can and cannot do this work through attrition and failure.

Your competitors are not waiting for the perfect assessment framework. They are running imperfect versions right now and iterating. Speed of honest evaluation beats precision of delayed evaluation every time.

This is not about replacing people with AI. It is about knowing which people can work with AI at the level your business requires -- and giving everyone else a fair chance to get there.