3 AImigo S1E2: AI Writes More Code. Why Isn’t Delivery Getting Faster? (#AI)

This Friday at 12:00 Moscow time, the new episode of 3 AImigo will examine a paradox that teams encounter more and more often: AI makes an individual engineer noticeably faster and produces more code, yet roughly the same number of changes reach users. Local speed is not the same as an accepted outcome, regardless of company size.

There will be three of us in the discussion: Evgeny Sergeev (LinkedIn), Alexey Litvinov (Telegram, YouTube), and me. We bring different perspectives: managing a large engineering organization, putting AI-Assisted Engineering into practice, and designing AI4SDLC architecture. We will not search for a single “correct” picture; we will compare our views.

We will discuss a product team where faster development sent more work into testing, but also increased returns and rework time. We will examine an AI-native startup whose agents complete many tasks, work in isolated environments, and take on-call duty after releases. We will also look at a large enterprise where AI tools are widely used, yet an important migration is still moving far more slowly than expected. And we will discuss the limits of automation: at a game development company, AI successfully handled migrations and changes that could be verified by compilation or tests.

We will talk about finding the real bottleneck and deciding what to optimize first; why fire-and-forget fails without a closed feedback loop; and how to build organizational memory by assembling context that agents can access, creating a semantic layer, and making knowledge genuinely searchable.

We will also discuss how engineering roles are changing: what developers now need to know, how the career ladder shifts, where responsibility moves, and whether hard skills really matter less. And we will ask why moving faster may be organizationally disadvantageous for some people and departments.

The episode will include not only success stories, but also candid examples of what did not work. Its central question is: what must change in the development system for agent speed to become business speed, rather than merely a new volume of code?

#AI #AI4SDLC #Agents #Engineering #Architecture #Management #Metrics #DevEx #Productivity #Software