Building Internal AI
Internal AI is everywhere now. Agents in Slack, agents reviewing PRs in GitHub, agents triaging support tickets, agents pulling spend data out of Ramp at 3am. The companies that figured out how to ship them are seeing wild numbers. Cloudflare's merge request velocity nearly doubled. Browserbase first-response times on support dropped under 24 hours. Sentry's Seer compresses what used to take an engineer half an hour into a single message.
The companies that didn't figure it out have a different problem on their hands. A sprawl of half-trusted agents acting on behalf of employees with overlapping permissions, talking to each other through APIs nobody fully audits.
After AI Council ends, we'll sit down with four leaders to talk about both sides of that.
David Cramer, Co-Founder and CPO at Sentry. Shipped Seer, which lets users ask questions about their app based on everything Sentry already knows.
Paul Klein IV, Founder and CEO at Browserbase. Built "bb", an internal agent now in daily use across engineering, ops, sales, support, and exec.
Brendan Irvine-Broque, Senior Director of Product at Cloudflare. Cloudflare's internal AI engineering stack now serves 47.95M requests across 295 teams, with 93% of R&D using it.
Ian Livingstone, Co-Founder and CEO at Keycard. Building the identity and authorization layer for agents acting on behalf of users without breaking the chain of trust.
Panel moderated by Allie Howe host of the Insecure Agents Podcast.
What we'll get into:
How to design agents work across the org in Slack, Linear, GitHub, Google Workspace, etc
The credential handoff problem: how leading teams keep agents from ever seeing real secrets, and what changes the moment one agent delegates to another
Shared agent memory without leaking one employee's context into another's session
Skill marketplaces, and how one engineer's breakthrough becomes the whole org's baseline
Adoption past the eng team. What actually gets sales and marketing teams using internal AI.
The numbers behind the hype. Which metrics moved a lot, and which ones surprisingly didn't
If you're building internal AI, securing it, or trying to get your org to adopt it, come hang out.