Talk Information
- Draft Title: AI-Native Teams: From Individual Productivity to Collective Advantage
- Length:
- Links:
- Targeting: June 25, 2026 event (https://lu.ma/ug32hek4)
Description
An engineer with Cursor or Claude Code can produce 3-5x more output in a day. Yet the team's throughput doesn't triple. Something absorbs the gains — and it's not the tooling.
This talk traces the evolution of AI-assisted engineering through three stages (autocomplete → imperative agentic → declarative agentic), and shows why the jump from individual productivity to team-level advantage requires a different kind of thinking: not better tools, but a different approach to coordination, ownership, and trust.
The core argument: declarative single-player mode is now mature enough to become multi-player. Agents can hold a declared responsibility over time, not just execute a task and return a result. That means teams can bring them on as narrow-focused contributors: "maintain code quality across all repositories," "track engagement and surface what matters," "manage our security posture." Giving agents declarative goals, the same way you lead a team.
The talk offers a practical four-level progression: from embedding Claude in a shared Slack channel, to scheduled intelligence agents, to agents that listen and act on repo changes, to custom team infrastructure, with concrete examples at each level. And it covers the leadership patterns that make adoption actually stick: framing goals instead of tools, making the agent's work visible, acknowledging what engineers feel as their role shifts.
Presented at the Agentics Foundation Vancouver meetup (May 2026).
About the Speaker
- Name: Anton Drukh
- Mini-bio: Anton Drukh coaches CTOs and VPs of Engineering at venture-backed companies from Seed through Series C. He was the first hire and VP Engineering at Snyk, scaling the engineering org from 1 to 150 people through the company's growth to an $8.5B valuation. He has spent the last 5 years coaching 100+ engineering leaders across the US, EU, and Israel on the leadership and org design challenges of scaling technical teams.
- Picture for slides: (We use this pictures to create promotional content and to present your talk in our socials.)
- Would you like help with your presentation? (Feedback on notes, content?)
- Feedback is always welcome, but I'm confident I can deliver the talk as is.
- Social media link(s): (twitter, website, linkedin, etc.)
- Do you agree to the [CNCF Code of Conduct](https://github.com/cncf/foundation/blob/master/code-of-conduct.md)?
Talk Information
Description
An engineer with Cursor or Claude Code can produce 3-5x more output in a day. Yet the team's throughput doesn't triple. Something absorbs the gains — and it's not the tooling.
This talk traces the evolution of AI-assisted engineering through three stages (autocomplete → imperative agentic → declarative agentic), and shows why the jump from individual productivity to team-level advantage requires a different kind of thinking: not better tools, but a different approach to coordination, ownership, and trust.
The core argument: declarative single-player mode is now mature enough to become multi-player. Agents can hold a declared responsibility over time, not just execute a task and return a result. That means teams can bring them on as narrow-focused contributors: "maintain code quality across all repositories," "track engagement and surface what matters," "manage our security posture." Giving agents declarative goals, the same way you lead a team.
The talk offers a practical four-level progression: from embedding Claude in a shared Slack channel, to scheduled intelligence agents, to agents that listen and act on repo changes, to custom team infrastructure, with concrete examples at each level. And it covers the leadership patterns that make adoption actually stick: framing goals instead of tools, making the agent's work visible, acknowledging what engineers feel as their role shifts.
Presented at the Agentics Foundation Vancouver meetup (May 2026).
About the Speaker