Engineering Management in the Age of Agents

The article argues that coding agents may soon handle more of the implementation work, but that will not remove the need for engineering managers. Their role remains in bridging business and technical needs, setting expectations, and handling incidents and difficult relationships.

It also says that writing code has never been the hardest part of software work. The harder parts are understanding the business, defining the right technical solution, and dealing with people. The author says these skills become even more valuable in an agentic world, and advises EMs to keep time for building things and to combine their management experience with current coding skills.

Reading notes

#

• Coding agents may reach the level of a mid-level engineer within a couple of years, including working across multiple repos and producing maintainable code.
• Even then, someone technical will still be needed to manage agents, create guidelines, and handle incidents.
• Writing code is presented as only one part of the job; solving customer problems is the real outcome of software engineering.
• Engineering managers bridge the technical and business sides by explaining delays, communicating incidents, learning what marketing and sales care about, joining customer conversations, and helping create roadmaps.
• The article says great engineers need to understand the business side, manage expectations, and define the right technical solution.
• Mentoring interns taught the author that technical work often needs clear context and careful definition of risks, measurements, scale, monitoring, and edge cases.
• Those same skills matter when working with LLM coding tools.
• People skills are framed as increasingly important because AI agents will not solve company problems involving relationships and collaboration.
• The closing advice is to keep time to build things and combine EM skills with up-to-date coding skills.