One of our former clients wanted to use as much AI as possible in their development process. At one point their leadership requested that we adapt repository structure, documentation location and workflows to make it easier for AI Agents to and automation tools using MCP servers to contribute to the code, peer review process etc.
Alla reached out to Andrey Adamovich of Extreme Automation, a long-term trusted partner and Team Topologies advodate to discuss how much teams should adapt to accomodate AI tools versus the other way around.
We started a collaboration podcast - Plan A, where we compare our notes on technology and leadership. Remember to subscribe to our channel!
Andrey has recently become a Team Topologies advocate and Alla called him to discuss whether AI will fundamentally change team topologies.
We covered a lot in this episode:
Can AI agents truly become team members as some suggest? Why do old problems like observability remain unsolved if AI is so powerful? Do agents get overwhelmed faster than average human team members? What does MIT research say about the effect of using ChatGPT on human brains? Why is J.A.R.V.I.S from the Marvel universe not a hero and what makes Vision a hero? Should we feed code written in human-centric programming languages to LLMs or invent a new language? Why is it still important that developers understand how computers work? What do scientists and specialist engineers say about giving AI autonomy? Can AI play facilitator roles? How can agents help temporarily bridge skill gaps? Why do LLMs struggle with Deluge, Latvian language, and welder interview transcripts? Why should governments train their own models? Do we expect more from AI agents than human team members? What are our biggest successes and pet peeves using AI? What must teams do to adapt to the biggest change since Covid brought remote work? Which developer jobs do we think are at risk?