WorkOS

Thank you to our speakers, to @stripe, @databricks, and @Cloudflare for sponsoring, and thank you to the 600+ builders who came out to init().

.@modal is the platform for production AI, giving teams inference, training, and sandboxes that scale from zero to thousands of GPUs in seconds. Powered by WorkOS. 🚀

Filter data by the user's permissions before the model reads it, not after. Our AI agent permissions checklist puts a test behind that control and every other one.

workos.com/blog/ai-agent-…

“The p(doom) kind of solves the permanent underclass problem”

put your muse in it!

and come tomorrow to hack on it

luma.com/developers-aft…

Hey init() attendees! Share your awesome badge hacks here! What cool things are you making it do? 👀

Auth MD - the paved path for agents to use your app.

My @aiDotEngineer workshop is up: "Lifestyles of the AI-Native" with @zackproser 🎥

Voice coding across agent sessions, goals + loops, worktrees, test gates, adversarial review, scheduled agents.

youtu.be/AgBAOXVxt4A

Conference badges keep getting better and better!

This one from @WorkOS’s init() conference

Most AuthKit requests that sound like they need a rebuild don’t. Skip a screen, make signup invite-only for one customer, turn off auto-join: often, it just takes the right parameter or dashboard setting.

Here’s how to customize your flow with hints, branding, and org-level settings.

workos.com/blog/customizi…

Big week ahead for SF Tech Week ⚡

• October 7 → init() at SFJAZZ
• October 8 → Developers After Dark at the Exploratorium

init(): luma.com/init-conf
Developers After Dark: luma.com/developers-aft…

See you there!

Special thanks to our sponsors, @databricks, @Cloudflare, and @stripe.

📣 @timgl, CEO at @posthog, at init().

October 7 · SFJAZZ

Register: luma.com/init-conf

The init() badge isn’t just swag. It’s a computer you can program.

Get yours at check-in, flashed with the conference schedule and ready to hack.

Register for init(): luma.com/init-conf
Explore the init() badge: workos.com/init/badge

A session from eight hours ago shouldn't be enough to revoke every API key. Step-up authentication requires a fresh sign-in right before the action that matters, GitHub's sudo mode, in your Next.js app.
workos.com/blog/step-up-a…

.@taylorotwell, @samlambert, and @grinich on one stage on what's actually working on their agent stacks, and what isn't. SF, October 20. Spots are limited: luma.com/laravel-43ty

.@Get_Writer turns enterprise data into governed, on-brand AI agents that execute real work — built with the security and compliance Fortune 500 teams demand. Powered by WorkOS. 🚀

Close out #SFTechWeek with vintage tech, live ESP32-S3 board hacking, and the run of the Exploratorium after dark.

Thanks to our co-sponsors @browserbase, @PlanetScale, and @laravel.

RSVP here: luma.com/developers-aft…

𝐁𝐮𝐢𝐥𝐝 𝐟𝐚𝐬𝐭𝐞𝐫.

An agent's permissions can't be static when its task is unknown until runtime. WorkOS Airlock evaluates every agent action against its intent, then allows it, denies it, or sends it to a human.

More on WorkOS Airlock: workos.com/airlock