London

June 28–29, 2027

New York

September 15–16, 2026

Berlin

November 9–10, 2026

The AI mistakes to leave in 2026

And the decisions that will shape your team’s next 12 months.

Dibyo Mukherjee and Priya Wadhwa

Date & time

17:00

Register for the panel discussion

Sign up to join this panel discussion, hosted in partnership with Chainguard.

This field is for validation purposes and should be left unchanged.

In January, a few assumptions still felt safe: that review would keep pace with the code, that juniors would keep learning by working alongside seniors, and that a known vulnerability could wait for the next sprint.

Now, AI agents can now write, run, and ship code autonomously. The catch: they’re operating in legacy environments with protections built for human speed. When an agent can open a PR, pull in a dependency, and deploy, traditional safeguards simply can’t keep up.

According to Chainguard’s research, 77% of security leaders expect an attacker to exploit a vulnerability at their organization within 12 months.

Vulnerabilities are next. The same tools that write your code make its weak spots easier to find, for your security team and for attackers.

In this panel, engineering leaders will explain how to give agents freedom to move fast without compromising security and discuss the decisions they’re making heading into 2027.

Key takeaways:

  • Stop review bottlenecks with trusted tests and clear sign-off ownership
  • Assign an owner to every known vulnerability and keep upgrade paths short
  • Let AI agents help harden your systems while allowing you to operate at machine speed
  • Avoid cognitive debt: engineers should be able to explain AI-assisted code, and juniors should still learn from people

panelists:

Priya Wadhwa

Chainguard
Director, Engineering
Dibyo Mukherjee

Dibyo Mukherjee

Adobe
Lead Software Engineer

Moderator: