Skip to content
back to talks
roundtable · Podlodka Techlead Crew

How AI Changes the Tech Lead’s Role

September 30, 2026 · Podlodka Techlead Crew · 1 h 33 min

Alexander Polomodov and Rafael Tonakanyan discuss the tech lead’s responsibilities as agents write code, propose solutions, and review changes. Hosts Igor Antonov and Andrey Kuleshov ask about AI economics, engineering expertise, team education, and the limits of automation. The points below summarize the recording; Alexander’s position prepared before the conversation is available separately.

Longread: my position before the roundtableWritten summary

Roundtable participants

The discussion at a glance

  1. 01Delegating work to an agent does not remove responsibility for architecture, quality, and operations. Tech leads need a verifiable process that does not make them the sole reviewer of a growing stream of changes.
  2. 02More tokens and code do not establish value. The discussion compares cost per task, human and agent time, delivery speed, and outcomes the business can recognize.
  3. 03Expertise appears in how a task is framed: domain language, acceptance criteria, and boundaries on changes. The panel offers no settled answer on how junior engineers should acquire that foundation.
  4. 04Requirements, architectural constraints, behavioral checks, and decision history matter more when code is frequently regenerated. Reviewing intent and agent activity instead of every code change remains a disputed proposal, not an agreed decision to abolish code review.
  5. 05Task size depends on model capability, verification, and workflow economics. Automatically fixing minor defects can produce a queue of patches the team has no time to accept.
  6. 06Faster development demands another look at priorities. Obsolete backlog items do not become valuable because an agent can implement them; tech leads must connect product framing with verification of outcomes.