New course: Build agents that respond to users with not only plaintext, but custom UIs like charts, forms, and whiteboards, generated on demand and displayed right in the chat. This short course is built in partnership with CopilotKit🪁 and taught by Atai Barkai, co-founder of CopilotKit. You'll learn three... approaches: Your agent can pick from custom components you build, like charts and forms. It can compose new layouts from a set of building blocks you provide, like rows, cards, and text. Or it can incorporate existing third-party apps, like a whiteboard or a calendar, right inside the conversation. Skills you’ll gain: - Build agents that render custom components like charts and forms on demand - Build an app where the agent and user collaborate on shared data, beyond just the chat window - Place third-party apps like maps, calendars, and whiteboards right in your interface Join and build agents that give users something to see and act on!show more

@CopilotKit It was so great to make this course with you all!

@CopilotKit @ataiiam This is huge! @ataiiam - CEO of @CopilotKit, teaching the world how to build the next generation of applications! 🔥

@CopilotKit @ataiiam About time! Another step closer to building agents that are actually useful coworkers 🔥

@CopilotKit @ataiiam Been fun working on this course with the Deep Learning team. Excited for everyone to give the course a try 😄

@CopilotKit @ataiiam Seeing a Agent UI course on deeplearning is the day i've been living for I'd say there are more nuances here that are worth deeper conversations - but thats probably another course

@CopilotKit @ataiiam Talk about a full-circle moment! In 2012, I was grinding through your online ML course. Today, your platform is teaching A2UI, a project my team and I built. Huge respect, Andrew—thank you for being such a foundational inspiration to so many of us! 🚀

@CopilotKit @ataiiam Real-time UI generation in agentic AI will significantly increase inference costs. Most underestimate infrastructure needs until deployment. We track these hidden cost curves.

@CopilotKit @ataiiam Been building an AI agent pipeline and the UI limitation is real - users want charts and forms, not paragraphs describing what a chart would show.

@CopilotKit @ataiiam Love it! This course will bring devs up to speed with generative UI and fullstack agents in just 1 hour. Amazing collab and great work from the @CopilotKit team

@CopilotKit @ataiiam Sounds like a game-changer for client communication! Custom UIs in chat could seriously cut down on those back-and-forth explanation emails. Excited to see what @ataiiam has built with @CopilotKit.

@CopilotKit @ataiiam This is the gap most enterprise AI teams are stuck in right now. They can build the reasoning layer but the output is always a wall of text. Dynamic UI generation turns agents from "smart chatbot" to actual workflow replacement.

The UI layer is where most agent deployments break down in practice. Operators and end users need to see what the agent is doing, not just read its output. Custom components that render inside the workflow change what's actually usable in a business context. This is the missing piece for enterprise deployment.

@CopilotKit @ataiiam Huge for the ecosystem. CopilotKit has been pushing the right problem: the UI layer isn't cosmetic - it's the difference between demos and deployable agents.

@CopilotKit @ataiiam Congratulations, @ataiiam!

@CopilotKit @ataiiam The agentic era does not just automate tasks. It rewires what a team actually looks like. Org structures will have to follow.

@CopilotKit @ataiiam Chat interfaces that spin up a chart or form on the fly? That's actually useful, unlike most AI agent demos.

@CopilotKit @ataiiam The missing piece is local context: the UI an agent generates in Lagos or Johannesburg should reflect payment rails, bandwidth, language mix and trust norms. Agents that ignore distribution realities will look clever but land badly.

@CopilotKit @ataiiam Embedding whiteboards and calendars directly in conversation is the real unlock here. Chat-only agents were always a stopgap. I use Neo for agent pipelines and moving to UI-first outputs completely changes how teams actually interact with agent results.

@CopilotKit @ataiiam That's fascinating! As a CPA who built an AI OCR, I see huge potential in dynamically generated UIs. Imagine custom financial reports appearing right in a chat based on user queries. Efficiency boost!

@CopilotKit @ataiiam the chat window is the new operating system

@CopilotKit @ataiiam custom UI is where agents stop feeling like chatbots. the interface should be generated around the task, not forced through one text box forever.

@CopilotKit @ataiiam Sounds fascinating! Dynamic UI generation in chat could be huge for explaining complex financial data or getting client input. Imagine real-time budget adjustments visualized instantly! #AI #accountin

@CopilotKit @ataiiam @AndrewYNg this is awesome! excited to see how it can make chat interactions way more dynamic. can't wait to try building with it!

@CopilotKit @ataiiam As AI is getting productized, what makes great products to stand out from okay products is the user experience,

@CopilotKit @ataiiam That's fascinating! As an accountant always drowning in spreadsheets, the idea of dynamically generated charts within a chat interface sounds like a huge time-saver for analysis & client communication

@CopilotKit @ataiiam coagents pushing live react components mid-conversation is the actual unlock. been wanting a proper walkthrough of this pattern

@CopilotKit @ataiiam This changes things. Interesting implications for the future of work.

@CopilotKit @ataiiam Generative UI is the only way AI agents become actually useful for enterprise. Plain text chat is too slow for complex tasks. CopilotKit getting Google and Microsoft to use their protocol is huge for standardisation. I want to see how this handles state management though.

@CopilotKit @ataiiam Banger. Text-only agents were just my warmup set.

@CopilotKit @ataiiam Couldn't agree more! Synthetic data generation could solve training bottlenecks.

@CopilotKit @ataiiam The big shift is that chat becomes an app runtime, not a text box. That matters for low-bandwidth markets too: a simple generated form beats a 20-message back-and-forth on mobile data.

@CopilotKit @ataiiam

@CopilotKit @ataiiam What prereqs would you suggest to understand and implement these course concepts in terms of coding/computer languages?

@CopilotKit @ataiiam Wow, that's fascinating! As a CPA in Tokyo who's seen AI OCR transform my work, I'm excited to see how dynamic UIs in chat can further streamline data interaction and reporting. Arigato!