Code of Leadership #84 materials: does a developer become a manager in the AI era? (#Management)

The recording and summary of Code of Leadership S2E26 from October 6, 2026 are ready. With Andrey Dmitriev, co-founder of JUG Ru Group, we discussed how engineering work changes when some execution can be delegated to agents: who chooses the task, organizes the work, and takes responsibility for the result?

An agent can diligently produce a report that nobody reads. Work continues, resources are consumed, and everyone forgets to check who still needs it. We returned to this question during the conversation: before accelerating something, it helps to decide whether it is worth continuing at all.

We discussed: 🔸 Product engineers and the limits of specialization Andrey emphasizes domain knowledge and the ability to understand users’ needs. Experience with a programming language helps, but nobody may bring you ready-made requirements for an unfamiliar product anymore. 🔸 Checking usefulness before automation Andrey described a manager who temporarily stopped distributing some reports; certain recipients did not even notice. That is a good reason to find the actual consumer before assigning another agent to the work. 🔸 An agent’s lifecycle Connect it, teach it the work, supervise it, and disconnect it in time. Andrey proposes this management analogy; we discuss permissions, outcome criteria, and support for the engineer. Giving someone a subscription does not create a process on its own. 🔸 Knowledge that must survive delegation I described an approach in which code can be rebuilt from intentions, architectural rules, verifiable behavior, and the reasons for decisions. Andrey reminded us about contradictions in documentation. Functions, security, and architectural compliance all need checking. 🔸 Software for one task and interfaces for agents I talked about a personal website for choosing a London neighborhood; Andrey described a bot that clarifies requirements and creates a website. We also discussed limitations: data freshness, access to sources, and the difficulty of explaining our own preferences.

Our predictions about the future of professions remain our assessments. The question for a manager is already practical: what value do they add when part of coordination can also be automated?

Materials for the episode: 📌 Episode page with timestamps 🎬 YouTube, VK Video 🎧 Podster, Yandex Music, Apple Podcasts 📝 Text summary

#CodeOfLeadership #Management #Leadership #AI4SDLC #Agents #Engineering