How engineering leaders can use AI to optimize performance

If there’s one area where most engineering teams are not making the most of AI, it’s team management.

Figuring out how to better manage engineers is often approached like more of an art than a science. Over the decades, engineering management has undoubtedly become more agile and data-driven, with automated data gathering improving performance. But in the past few months, the evolution of AI — specifically, predictive AI — has thrown management processes into a new era.

Predictive AI analyzes data to foresee possible future patterns and behaviors. It can automatically set goals based on real-time data, generate recommendations for improving teams’ performance, and process far more information than was possible before.

I want to encourage all other engineering management and intelligence platforms to start using AI, so we can collectively move into a new era. No business wants to lose profits or market share because of bad management.

We now have the data and the technology to turn engineering management from an art into a science. This is how engineering leaders can use AI to manage their teams and achieve more with less.

Pinpoint hidden patterns

Even the most capable engineering leaders have some blind spots when it comes to reviewing performance in certain areas, and may miss concerning behaviors or causal factors. One of the most significant ways engineering managers can apply AI to their workflow is by generating full reports on engineers’ performance. Typically, managers will manually put together reports at the end of the month or quarter, but often that gives a superficial analysis that can easily conceal hidden or incipient problems.

In the past few months, the evolution of AI — specifically, predictive AI — has thrown management processes into a new era.

Predictive AI can automate insightful performance reports telling leaders where they should be making improvements. The main advantage here is that AI has a greater ability to identify patterns. It can process all existing data on a team’s performance, as well as internal and external benchmark data, to produce a level of analysis that humans can hardly attain at scale.

For example, AI can better analyze the relationship between cycle time, code review time, and code churn (the frequency with which code is modified). It can determine if longer code review times are actually leading to less code churn — which could imply more stable and well-thought-out code. Or, it may find that longer review times are simply delaying the development process without any significant reduction in churn.

By analyzing multiple metrics simultaneously, AI can help identify patterns and correlations that might not be immediately apparent to managers, enabling organizations to make more informed decisions to optimize their software development processes.

Another advantage is that AI tools can produce simple but analytical reports every day with 0 manual input, allowing managers and leaders to detect any important shifts in real time, not just at the end of every sprint.

Permanent memory bank

AI tools have a permanent memory of the progress of the team and company. Imagine what happens when an engineering manager leaves a business. Yes, the team’s performance data remains, but the wealth of knowledge that the manager has accumulated disappears. (Under what conditions does the team perform best? Were there external factors impacting poor performance? What strategies have been implemented and which worked best?)

For the first time, predictive AI can actually learn exactly what your team’s process has been so far. It can capture all that historical knowledge internally for your company, baking in that level of complex reasoning that can then be used by successive managers and future decision-making.

Maintaining a permanent data store of a company’s progress means key strategic info doesn’t get lost with staff turnover. It allows for a more fair assessment of the team and saves time and resources being spent on tactics that have proven unsuccessful.

Generate goals, targets and advice

Consider how predictive AI tools can act as a co-pilot to leaders. When they capture all the team’s internal data, they can turn it into equally unique goals and milestones.

Predictive AI tools can set goals for a team based on real-time data — for example, by automatically creating targets for the team on a weekly basis based on changes in performance. More importantly, they can come with built-in advice and use cases on reaching those targets. For example, a tool can identify a need to decrease cycle time, then set a target at 20% reduction, and offer a 12-month plan with advice on how to get there, with tips on how to improve handoff during product review, and so on.

These tools won’t just be wiring questions to ChatGPT and spouting unverified recommendations. They can be trained with input from experts that include advice, proven solutions, best practices and case studies. Engineering managers and management platforms have a wealth of internal and industry data to determine which approaches work best in particular conditions.

Of course, there are no cookie-cutter solutions. But anyone who has tinkered with predictive AI knows that it is uniquely capable of providing advice with a granularity that can take an unprecedented number of variables into consideration in every output.

At least to start, these tools will be a work in progress as teams train it to output more accurate and effective recommendations. Managers can focus their efforts on refining the tool’s output, or adjusting when necessary — for example, if it stops providing the desired results, or if internal/external conditions change and warrant a shift in strategy.

Two-factor verification

The subjective nature of managing a team can be hard for engineering leaders. Often, they’ll perceive that something is wrong but can’t find any proof of it. Or they’ll spot changes in performance but won’t be able to pinpoint the reasons behind it.

Predictive AI can be a sort of “two-factor verification” for engineering leaders to validate their intuition based on data. Because the technology is able to process more unstructured data and prompts when analyzing information, it can dig up causal factors that are imperceptible to the human eye.

For example, if an engineering team is having to deal with an unhealthy number of bugs in code, but all their metrics are hitting general benchmarks, a manager may not get much insight from the data as to why. But predictive AI can make a connection between metrics in order to provide solutions and advice. For example, it may connect a high deployment frequency as metric A and the high speed of the review stage of the cycle time as metric B and determine that the team is not spending enough time reviewing code, which is letting bugs through.

Predictive AI can also allow engineering leaders to play out certain scenarios to identify ideal paths forward. They may be contemplating if a team would do better if they hired an extra developer versus another approach, such as redistributing workload. With the right data, AI can run those scenarios in minutes and suggest possible outcomes so that managers can make an informed decision.

It’s important that engineering leaders always keep in mind that the human “variables” are still their responsibility and that some aren’t automatically weighted by AI. Developer experience and well-being may not be tangible in certain metrics, so make sure you always bring that balance to your considerations when using AI tools.

Technology follows the path of least resistance, and engineering leaders always opt for optimization. While some fear they will lose their jobs to AI, I feel like this evolution will instead adapt jobs to today’s world: a world in which tech workers will have to learn to use AI to better achieve goals. That’s why I invite all forward-thinking managers to explore the potential of AI as a complementary resource to elevate their development processes.