Oso

Today Webflow launched CMS Collection access control for Enterprise customers 🚀 Get the behind-the-scenes story on how they partnered with Oso to move beyond a homegrown model and deliver fine-grained permissions at enterprise scale.

Join us for an exclusive dinner during AWS re:Invent Hosted by Oso, at the Encore Las Vegas, this private dinner brings together engineering leaders attending AWS re:Invent for an evening of conversation on AI, agent systems, and the security challenges shaping this new era. Space is limited — request your invite to be considered.

What is different about identity for agents versus traditional identity? A lot. And we’re still very early in figuring all of this out. We’ve had a recent run of funding and acquisitions for LLM security. That’s no surprise – LLM adoption happened rapidly, so we had to secure the models and user interactions with them first. Agents are getting better quickly, though. It might be a struggle to find production use cases today, but that won't be the case a year from now. We’re in the middle of a major platform transition with AI. ChatGPT (reportedly) has 800 million weekly active users and 2.5 billion prompts per day. Platform shifts create net new capabilities and put the traditional center of gravity in tech up for grabs. OpenAI just announced apps in ChatGPT and a new browser. Claude Code runs on the web now. We see major announcements like these every week. Meanwhile, there are still tons of industry-defining questions nobody knows the answer to yet: – Will everything be built around clients like ChatGPT and Claude? – Will AI be more integrated into apps and agents like Notion and ServiceNow? – Will things converge around something else, like browsers (or a whole new device?) From a security standpoint, I’m sure of one thing: it’s complex to find the right solutions and controls right now. The lines between established and emerging security categories are getting blurred while everyone figures out how to handle agentic use cases. I’ve been busy trying to sort out this market in a way that’s practical and usable. There’s a lot of good research out there already, and I’ve been consuming a ton of it. The team at Oso has some of the best thinking I've seen on this topic. That’s why I'm excited to collaborate with them on some research to map out this emerging market. This feels like some of the most significant research I’ve ever done. Teams shipping AI agents aren’t just adding a model — they’re introducing autonomous actors that interpret intent, chain tools, and take real actions on your behalf. I’ve never experienced anything that’s so powerful and so risky at the same time. The AI platform shift moves us way beyond the controls we’ve historically relied upon for traditional apps. This market map is your guide to the technology stack that comes next. We’re going to work through the main use cases, discuss where the boundaries are today, and share places we’re seeing overlap between new and existing identity solutions. More posts soon!

OAuth is a broadly accepted standard. It’s used all over the internet. But for LLMs and agents, OAuth isn't enough. https://lnkd.in/e6Hai__t

We’re hosting an exclusive dinner and discussion on AI and agent permissions. If you’re in the Boston area, join Oso and a select group of engineering leaders to explore how teams are approaching AI, agent systems, and the security challenges shaping this new era. RSVP to join the list.

Today, we’re publishing a guide on building an authorized RAG chatbot. WTH is that? If you’re adding an LLM-powered chatbot to your app, you don’t want it serving up just any data—it needs to respect permissions, just like the rest of your app. How do you do this? Here’s a technical guide on how to solve this problem with Oso, Supabase, and OpenAI. Read the post for more info: https://lnkd.in/gWZvujbP