Skip to content
News Figma Blog Aug 2026

Figma Sightlines: what leaders need to know about building AI agents

Figma’s Sightlines newsletter published its second issue on August 20, 2026, focused on agentic products. The central question is practical: how do you build a tool capable enough to be genuinely useful while keeping trust and security intact?

The issue draws from executives at Cisco, OpenAI, Airbnb, Atlassian, Accor, and Expedia. Three themes run through their contributions.

The first is transparency as a prerequisite, not a feature. Thomas Vidal, VP at Accor, frames the goal not as designing what AI can do, but as creating conditions for consistent system behavior. For users, this means agents that show their work — citing sources, narrating decisions, communicating what they are about to do before they do it. Leaders across the issue push product teams to build this in from the outset, not retrofit it after launch.

The second is accepting that launch will be imperfect. Rachel Been, SVP at Expedia, argues that perfection at launch is unrealistic: real-world usage provides learning that controlled testing cannot. The recommendation is to ship once baseline trust requirements are met, then iterate based on actual behavior. This stands against a tendency in enterprise software to delay releases until every edge case is handled.

The third is governance scope. Charlie Sutton, Chief Design Officer at Atlassian, raises the accountability question: as agents handle more decisions independently, teams need alignment on which tasks agents can run autonomously versus which require human review, particularly for consequential actions. Getting that boundary defined early avoids both over-delegation and the friction of an under-trusted system.

Relevant for design and product leaders building agentic products or managing teams that are. The newsletter is free and aimed at senior practitioners rather than individual contributors.