Building the AI-Native Organization

From Density 2026: Tigran Sloyan (CodeSignal), Dan Pupius (The General Partnership) and Maria Kazandjieva (Figma) on how AI is reshaping engineering teams, managers and who thrives.

Aki Ito, a reporter at Business Insider, moderated a panel with Tigran Sloyan, co-founder and CEO of CodeSignal; Dan Pupius of The General Partnership, which embeds with early-stage startups; and Maria Kazandjieva, an engineering leader at Figma. Her premise: software engineering changed first, so it shows where every other job is heading.

Speed can cost focus. Dan said a single engineer now has more leverage and launches come faster, but founders are tempted to sprawl. Maria said Figma has kept teams lean and is retooling developer experience and internal tools so people and agents can work in the same code base.

Jobs aren’t changing. The path is. Tigran defines a job as inputs, outputs and the work in between. Engineers still deliver features. What’s changing is how they get there and the skills that takes. He doesn’t expect fewer engineers overall: when work gets cheaper, companies want more of it.

Engineers are becoming managers of agents. Maria sees more engineering time spent breaking down work and coordinating agents, and some engineers going through the overwhelm of a new manager. Some middle-management work is moving down to them, which is part of why org charts are flattening.

Openness predicts who thrives. Tigran said generalists high on openness adapt best, and the strongest assessments now simulate messy work, like probing an unhelpful AI product manager before analyzing data. Maria finds senior generalists who understand architecture get the most from AI, while new grads learn faster than ever.

The overwhelm is real. Dan described the compulsion of “one more prompt” and the tax of watching many threads at once. Tigran called it temporary: today’s agents act like eager junior employees who need constant direction.

Their advice: own your upskilling (Tigran), learn how to learn (Dan), and sharpen critical thinking and systems knowledge, including what makes a code base good for agents (Maria).

Related Articles