Why Engineering Managers Remain Crucial in the Age of AI Agents

It seems we are entering the era of AI agents, but I remain optimistic about engineering management.

We can assume that future programmers, after a few years of learning, will reach a true mid‑level engineer status—understanding multiple codebases, avoiding chaos, and writing maintainable code.

Even then, developers will still be needed to manage these AI agents, set guidelines, and handle incidents when they arise.

In practice, writing code is never the hardest part of our work.

The outcome of a software engineering organization is not merely “runnable code”; the most important thing is solving customers’ problems—something we never learned in university.

“A doctor once told me the most important thing medical school doesn’t teach is the difference between medicine and being a doctor.” Medicine is a biological science, while being a doctor is largely about managing expectations, understanding insurance systems, and effective communication.

The software development world is similar. Most software engineers receive training in computer science, but becoming an excellent engineer also requires business understanding, architectural design, managing expectations, and defining the right technical solutions.

1. Understand All Aspects of the Business

An engineering manager acts as a bridge between technology and business, explaining project delays, communicating important events, understanding marketing and sales concerns, and engaging in dialogue with customers.

Explain why projects are delayed.

Convey important activities and events.

Understand what marketing and sales focus on.

Engage in conversations with customers.

Help define the R&D roadmap.

To do these, we must deeply understand the business: how we make money, what customers care about, and how our product is marketed. These conversations happen before any code is written.

2. Explain Your Requirements

I have mentored many fresh graduates. Initially, you give them small, concrete tasks and plenty of system background; gradually they become familiar and can handle more complex, ambiguous work.

Even when managing mid‑level or senior engineers, you often need to define or help define technical solutions, including:

What risks are we willing to take?

How will we measure success precisely?

What scale are we aiming for?

How will we monitor it?

Which edge cases must we support?

3. Deal With People

Engineers usually dislike “people problems” such as annoying product managers, unresponsive UX designers, or difficult teammates. Only a few engineers proactively hold one‑on‑one meetings, share feedback, and try to resolve issues; most ignore the problem or ask a manager to intervene.

In the upcoming “agent” world, interpersonal skills will become even more critical. AI agents won’t solve all company problems—companies are always made up of people.

Mastering these skills—being able to handle complex interpersonal dynamics and enjoy collaborating—will be a huge advantage.

Conclusion

I know being a project manager or engineering manager isn’t as easy as described here, but I believe it’s an exciting opportunity. Your current skill set may be exactly what companies are looking for.

My advice: ensure you have time to build products. If you can apply all the skills you learned as an engineer and combine them with the latest programming capabilities, you will be in high demand.