Discussion – GenAI and dev platforms – how development is changing (AI column)
At the summer IT Picnic, my colleague Dima Gaevsky discussion on this subject, which was attended by distinguished gentlemen.
- Ivan Puzyrevsky, CTO Yandex Cloud Alexander Lukyanchenko, Head of Platform Development at Avito Stanislav Sychev, Head of Platform Development at T-Bank
The students discussed the following topics 1. The role of the developer is changing AI is no longer just part of the IDE. AI writes code, offers optimizations, generates tests – and the engineer becomes a reviewer and solution architect, not just an executor. Strong engineers now oversee both humans and AI agents. The Junes grow faster - they have an "AI assistant" at hand, which tells them how to write code. The main skill of the future is to be able to set AI tasks and check the results, not just know the syntax of language. 2. AI becomes part of development platforms
- Platforms. (CI/CD, IDE, clouds) integrate AI assistants that analyze code, collect metrics, offer fixes. AI models learn from internal repositories, understand corporate code style and standards. As a result, internal dev platforms are becoming “living systems” where AI not only helps people, but also learns from their patterns. 3. The autonomy of AI is not so great yet. No one believes in an AI developer without a human. Now the boundary is clear: AI can be trusted with routine (generation of templates, tests, fixes)But not architecture and production code. Everywhere there is a rule: “AI offers, engineer approves” (There is a human in the loop.). It has to do with the cost of mistakes. 4. New roles and processes New professions appear: AI-engineers, prompt designers, model reviewers. Some partially deterministic scenarios work better: generation of test cases, unit tests, code review agents, code migration DevEx platforms are transforming into ecosystems for people and agents: both learn from shared data and increase efficiency. 5. What next? AI-first platforms will become the standard: without the built-in assistant IDE or CI/CD soon will not do. Teams will become smaller and more interdisciplinary – machines take routine, people are left with design, logic and responsibility for their work and the work of AI agents. AI literacy will become the new basic competence of the engineer. Agent environments will appear, where several AI services will solve the problem “from TK to Deploy” under the supervision of a human teamlid.
If you're going to sum up, guys. They don’t believe that AI won’t replace engineers; rather, engineers will become teammates of AI agents. Development platforms become smarter by integrating AI capabilities into themselves
#Software #Engineering #Productivity #DevEx #AI #Management #RnD #Leadership #Economy