We've all seen the headlines. "AI will replace coders." "The end of engineering as we know it." It’s a narrative designed to grab attention, but I don't buy it. It doesn't help anyone, least of all the brilliant, collaborative, problem-solving human beings who make up our engineering teams. For all the hype you see, how many concrete examples of workflow improvements can you think of?

My time as an engineering leader has taught me one absolute truth: software development is a team sport. It thrives on connection, communication, and human intuition. The question we should be asking isn't, "How do we replace engineers with AI?" but rather, "How do we use AI to supercharge our teams?"

By taking a people-focused approach using the CAMP (Connection, Autonomy, Measurement, Purpose) framework, we can win with AI and build stronger teams. Here’s how focusing on human strengths, guided by key leadership principles, unlocks incredible engineering outcomes in velocity, quality, collaboration, and innovation.

CONNECTION: AI is Not Your Team Member

Let's start with a crucial distinction: AI isn’t actually creative. It can't feel the thrill of a breakthrough, the frustration of a stubborn bug, or the satisfaction of a seamless production deployment. It can generate code, analyze data, and suggest patterns, but it cannot empathize or understand the context that drives true user connection.

The real innovation happens when we recognize that people can do creative things with AI. AI excels at repetitive, predictable, and boilerplate-heavy tasks. When your engineers use AI to automate their repetitive workload, they free up their time to tackle the more interesting problems they signed up to solve. It frees up mental bandwidth for the innovative work that requires genuine connection and collaborative brainstorming.

I’ve seen teams use AI as the first step in a pull request review. The AI made several suggestions on how to improve the code. This gave the team topics to discuss and debate. This strengthened the team as they learned more about each other's views on different areas of their craft. Some of the suggestions could be thrown out because the AI didn’t know the patterns to be used or because it hallucinated. The team was able to focus on real improvements to their code.

AUTONOMY: Trust Your Devs to Choose Their Tools

The fastest way to kill morale is to force AI tools where they don't fit. Don’t mandate that AI must be used everywhere. A blanket requirement that every pull request include an AI contribution is micromanagement.

Instead, win by providing Autonomy. Put AI tools in your engineers’ hands and trust them to use AI where it’s genuinely useful. They will figure out that it’s fantastic for generating unit tests for edge cases or drafting initial boilerplate code. They will also realize where it fails, which, it turns out, is surprisingly frequent. This autonomy fosters ownership and encourages engineers to become discerning judges of quality, learning when to rely on a tool and when to lean on their own expertise.

MEASUREMENT: Focus on Real Impact, Not Just Speed

There’s a lot of noise about AI dramatically speeding up development. Is that true? AI can certainly write many lines of code quickly. Is it the right code, though? Instead of guessing, we need practical measurement. A simple experiment to run: Have one small team explicitly trial an AI tool for a few sprints, while a similar team in the same domain does not. After the set number of sprints, does cycle time go down while bugs stay low?

Compare the results. Does the AI-augmented team actually move faster (shorter cycle time)? Is their code quality better or worse? (Think defect rate). The data might surprise you. Some studies suggest that AI performs poorly compared to a human, with an astonishing 97% of the time on complex programming tasks. Reports indicate that 95% of companies are not yet seeing measurable value from their AI initiatives. Your job as a leader is to measure, not just assume, the real impact AI has on your unique engineering outcomes.

PURPOSE: A Better Future, Not an Unemployment Line

This brings us to the real opportunity. We can't let the primary narrative around AI in engineering be one of replacement or cost-cutting. That’s a recipe for burnout and anxiety.

The most vital shift is changing the narrative to Augmentation with Purpose: Opportunity, not Mandate. When you help your team automate something they regularly do, the parts of the job that feel repetitive or soul-crushing, the toil, you aren't taking away their value. You are letting them focus on something more interesting. This is the definition of growth. It’s the difference between doing grunt work and designing systems, solving problems, and delivering genuine value. That is a future worth winning.

Practical Takeaways for Engineering Leaders to Start Now

You don't need a massive strategy or a new department to start balancing AI with human strengths. Here are some small experiments you can implement:

1. 1. Run the "A/B Sprint" Test: Select a team willing to be guinea pigs. Provide them access to a specific AI coding assistant (like Copilot or a code-generation LLM). Dedicate a sprint or two, letting them use it for unit test generation or boilerplate creation only. Contrast their velocity and code quality metrics against a matched control team working in the same codebase. What do the real numbers tell you? You'll know AI is having the impact you want when velocity increases, and code quality remains at least the same.

2. 2. "Report What's Been Automated" (For Accountability): Introduce a small reporting loop. When an engineer finishes a substantial task, ask them a simple, optional question during their sync: "Did you use AI to help automate any part of this work (code, tests, docs, etc.)? If so, which part?" This creates visibility, helps identify repeatable opportunities, and fosters a culture of shared knowledge about what’s actually working. Your team will start naturally sharing ways to improve how they work with AI.

3. 3. Start the Purpose Conversation: In your next team meeting, explicitly ask, "If you could instantly automate 2 hours of your most repetitive work, what interesting problem would you spend that time on instead?" This immediately shifts the focus from anxiety to opportunity and growth, aligning AI use directly with individual motivation. When your team is excited about what they can work on, they’ll deliver value you don’t know to ask for and won’t expect.

4. 4. Define what success looks like: Align with your team on how you’re going to measure the success of your AI initiatives. The measurement should be outcome-based: “We want to deliver high-quality features faster as measured by our cycle time and the number of bug reports we see from the field.” Don’t focus on output metrics like lines of code. AI can produce a lot of code really fast, but that’s not what your customers need. Identifying the outcomes you're chasing will give your team the freedom to suggest new ways of achieving them. When your team has the freedom to forge their own path, you’ll see them engaging like real owners.

These experiments map back to CAMP, they are a way to build connection, give your team autonomy, ground decisions in measurement, and point the work toward purpose.