Web Summit Vancouver 2026 brought some of the most important conversations in AI to the stage, and 1Password CTO Nancy Wang was leading them.
On Centre Stage alongside Genspark co-founder and COO Wen S., and Emil Protalinski, Editor at Techmeme, Nancy drew a sharp line through the current moment: in under 12 months, organizations went from blocking ChatGPT entirely to handing agents near-godlike admin access, with almost nothing in between. The missing piece isn't more tooling. It's knowing how to give agents exactly what they need and nothing more.
And the framing matters: governance and productivity aren't at odds. Bad governance is. Get access design wrong and you're spending months in incident reviews. Get it right, and least-privilege becomes the thing that lets you scale.
"Everyone talks about, 'I want autonomous agents.' I don't think they do. I think they want obedient interns with perfect memory and speed."
At the AI Summit, Nancy joined Jennifer Smith, Co-Founder & CEO at Scribe, Linda Tong, CEO at Webflow, and John Koetsier, columnist at Forbes, to address the gap between model performance and what organizations are actually getting. The bottleneck isn't the models. It's the missing context layer, the governance gaps, and identity challenges that traditional security frameworks weren't built for.
"Agents aren't just machines. They can reason. They're non-deterministic. And so even if they have the right access, the question now becomes: are they actually going to do the right thing? Agents need their own identity and a way to bind intent to that identity so they can be monitored as they get access to sensitive systems."
The thread connecting both conversations: trust isn't a feature you add later. It's the architecture.