AI Native DevCon’26: The London conference for developers building with AI

The bottleneck moved from writing code to governing it.

The promise was 2× throughput. The reality is 2× the review queue, 2× the security exposure, and a CI signal you can no longer trust. AI Native DevCon 2026 is for the engineering leaders who have to figure out how to ship anyway.

AI Native DevCon 2026 lands at The Brewery in London on June 1 and 2, with a hybrid track for remote. This is the conference for VPs of engineering, CTOs, platform owners, security leads, and senior engineers running agents in production, or about to. 500+ builders. Four tracks.

Guy Podjarny, founder of Tessl, organizer of AI Native DevCon, and previously of Snyk, frames the 2026 question:

“If 2025 was the year coding agents started showing real promise, 2026 is the year we figure out how they hold up in production. The challenge is no longer getting an agent to work, it is getting it to work consistently across teams, codebases, and environments without constant human correction.”

The schedule is organized around four tracks: context engineering (building with agents), agent orchestration (verification when CI is no longer enough), organizational enablement (coordination at agent throughput), and agent enablement (security and governance). Each maps to a problem most teams are already hitting.

The agenda is built around the problems they actually have right now.

1. “We do not know how to build with agents yet.”

How the engineer’s role is changing, and what products designed for humans need to do once agents start using them. By 2026, that question lands on every platform team. This is the context engineering track.

• Ryan Lopopolo (OpenAI), Harness Engineering. Concrete patterns for systems where humans set direction and agents execute, including the review and approval surfaces that keep it safe at scale.
• Dana Lawson (Netlify, CTO), Built for Humans. Now Agents Are Here. What changes in a developer platform when half the users are non-human, and the API and UX decisions Netlify made in response.
• Ian Thomas (Meta), AI Native Engineering. How a large engineering org is restructuring workflows around agent-assisted development.
• Steve Ruiz (tldraw), Agents on the canvas. Interaction patterns for visual agents, with shipping examples you can copy.

2. “We can generate code. We cannot verify it.”

CI is no longer evidence of correctness. Two years of agent-generated code has proved it. The agent orchestration track is about what to put in its place.

• Justin Cormack (ex-Docker CTO), When Tests Lie. Runtime signals that flag agent-introduced drift before it reaches users.
• Dave Farley (Founder & CEO of Continuous Delivery Ltd. - 250k on Youtube), Vibe Coding, really? The ideas that may actually survive the AI programming revolution, beyond hype, demos, and generated boilerplate.
• Chad Fowler, Regenerative Software. An architectural model where components are regenerated rather than patched, and what verification looks like when code is short-lived by design.

3. “AI writes code faster than teams can coordinate.”

Two years into coding-agent adoption, throughput is up roughly 2×. Coordination cost scaled with it. The organizational enablement track covers review, ownership, and team structure.

• Guy Podjarny (Tessl), Skills are the new Code (keynote). The case for treating skills as proper software: versioned, tested, owned, reviewed. With the Tessl Registry now holding 2,000+ evaluated skills, the talk covers what that means for repo structure and review process.
• Birgitta Böckeler (Thoughtworks), State of Play: AI Coding Assistants (keynote). Two years of field data on which adoption patterns work and which create future technical debt.
• Patrick Debois, The Rise of Agent Enablement. Agent Enablement is the function that owns reliable agent adoption inside an engineering org. It defines standards for skills, evals, and workflows, and sits next to DevOps and Platform Engineering. Patrick’s session covers who owns it, what they do, and how teams formalize it.

4. “Your AI is a new attack surface.”

Vulnerability classes that did not exist 18 months ago, and the controls most teams have not put in place yet. This is the agent enablement track from a security and governance angle.

• Liran Tal (Snyk), Your AI Agent Installed Malware Because a SKILL.md Told It To. Live demo of prompt-injection via SKILL.md manifests, with the threat model and mitigations.
• Joseph Katsioloudes (GitHub), Code Security Reinvented. How SAST, secret scanning, and review need to change for AI-generated code.
• Jack Wotherspoon (Google), Humans vs. Slop. New rules for open source maintainers when an unknown share of contributors are agents.

Why engineering leaders should attend

Five things your team brings back to Monday:

1. A verification model that does not assume CI catches the regression
2. Threat models for prompt-injection and SKILL.md attacks, with mitigations
3. Team structures and review workflows that scale with agent throughput
4. A working definition of Agent Enablement as a discipline, including ownership and scope
5. A model for evaluating skills before they go org-wide, with review patterns and KPIs

Hosts and the wider lineup

Hosted by Sam Hepburn and Patrick Debois. Day-one keynote from Lieven Scheire on AI from outside the engineering bubble. The wider roster covers agent observability, MCP transports, runtime intelligence, brownfield adoption, and team-level adoption metrics, with practitioners from Anthropic, OpenAI, NVIDIA, Adobe, Hugging Face, Mozilla.ai, Cisco, Nearform, GitHub, and much more.

Full speaker list and abstracts: tessl.io/devcon.

TL;DR

• Dates: June 1 and 2, 2026
• Format: 2 days in-person or 1 day virtual
• Venue: The Brewery, Barbican, London.
• Register: https://luma.com/aidevcon-ldn26?coupon=R30 (R30 auto-applies at checkout to knocks off 30% off)
• Bringing a team? Contact at tessl.io/get-in-touch, and we can arrange a group purchase discount.

See you at the event!