.jpg)
Part 2 of 2. Part 1 was about where the engineering bottleneck moved. Part 2 is about the career ladder underneath it. AI is changing how engineers enter the field, how they grow and what great leadership looks like. The leaders we spoke with on Day 2 weren’t waiting for answers. They were already redesigning their teams.
As in Part 1, attendees are described by role rather than name, and their questions are condensed.
Q — Engineering leader, growing software company: We used to bring on four or five junior engineers a year. AI now covers much of that work, so we’ve paused junior hiring. But our senior engineers of 2030 have to come from somewhere. How should we think about that?
It was one of the most thoughtful questions we heard all week, and a common one. Students graduating now are arriving with AI-native skills, and new roles are appearing fast. As he put it, data scientist wasn’t a job title when he graduated either. The opportunity is to redesign what “junior” means.
Lucas Hendrich (CTO at Forte Group): You’re no longer hiring for a programming language. A strong engineer who has built payment systems in Java can rebuild them in Go, with AI handling the syntax. What matters most is domain judgment: understanding why a system works the way it does. That’s what early-career engineers can start building from day one.
Our take: AI excels at tasks, which frees early-career engineers to grow into decision-making sooner. Teams that pair juniors with AI, and with a senior mentor, can develop judgment faster than the old ticket-by-ticket path ever did.
Monday move: Ask your engineering managers: “Who are we developing into our senior engineers for 2030, and how?” It’s a quick way to make long-term talent planning part of this year’s roadmap.
Q — Product and engineering leader: My strengths are business priorities and org design. I’m less hands-on with AI than some of my peers. How should leaders like me approach this?
It’s a question many experienced leaders are asking, and asking it is a strength. Leadership in the AI era isn’t about writing prompts. It’s about designing how people and agents work together.
Lucas: Agents are great at tasks; decisions still belong to people. Rather than putting an agent in a box on the org chart, look at each role and separate the tasks that can be automated from the decisions that need human judgment. That’s a leadership skill, and it builds directly on org design experience. The key is understanding the mechanics well enough to know where agents need guardrails.
Monday move: Choose one team and map its weekly work into two columns: tasks and decisions. Tasks become your automation backlog. Decisions are where your people can create even more value.
Q — Senior engineering leader: Public messaging makes it sound like everyone has AI fully figured out. In one-on-one conversations, people are much more open about what they’re still learning. What should we take from that?
Lucas: That this is how a field moves forward. The most valuable conversations at ELC are the candid ones, where leaders compare notes on what’s actually working. And there’s real progress to build on: shared patterns are emerging, including skills, playbooks, harnesses and MCP. It’s a lot like the early days of modern web frameworks. The foundations are forming, and they’ll keep getting better.
Our take: Build with flexibility in mind. Design your AI stack so models, harnesses and vendors can evolve without a rewrite. Anchor every initiative on a business KPI rather than a specific tool, so the value lasts as the tools get better.
Monday move: For each AI tool or framework in production, write down how you’d swap it if something better came along. It’s a quick way to keep your options open.
An interesting view from the floor: as software interfaces become more “headless,” buyers may come to prefer transparent, usage-based pricing such as tokens over traditional subscriptions. That shift could change how AI products are built and sold.
83% faster delivery. Under 10% rework. See exactly how Xceptor got there.
%20(1).jpg)
