CopilotKit🪁

⭐️Introducing AG-UI; The Agent-User Interaction protocol 👾 Bring your AI agents into Frontend applications, and let them interact with users. Launching with day-0 integrations with LangGraph, CrewAI, Mastra, and AG2 - with more partnerships on the way 👀 github.com/ag-ui-protocol… How it works & demo 👇

Any AI agent can now do what Claude does! I found an open-source version of Anthropic's new Generative UI tools for Claude, built using @CopilotKit. It lets you render live charts, interactive diagrams, and 3D scenes right inside your own AI apps, just like Claude. Check this:

@googledevs AG-UI 🤝A2UI

Introducing the Developer’s Guide to AI Agent Protocols. Stop writing custom integration glue for every tool, API, and frontend. We used Google Agent Development Kit to build a B2B agent that handles the full stack using 6 open standards: MCP, A2A, UCP, AP2, A2UI, and AG-UI The future of agents is interoperable. Read the full technical guide here: goo.gle/4sRrgl2

AG-UI mentioned 🩵

CopilotKit-powered Excalidraw MCP app! I'm using the @CopilotKit - @excalidraw studio app in my own workflow. Shoutout to @Anmol_Codes for building it!

What do you think @excalidraw 👀

Bring agent interactivity and generative UI into your own app: github.com/CopilotKit/Cop…

Introducing @Excalidraw Studio: Generate real, editable diagrams as MCP Apps inside your chat Edit in local canvas and refine via chat all within the same session. Auto-save to disk with persistent workspaces, no database. Then push it live in one click. Open-sourced for the community. 👾 Repo: github.com/CopilotKit/exc…

I built a retro game with a Pac-Man-style theme. 100% @CopilotKit Headless UI and zero prebuilt chat components. @tavilyai under the hood - that searches the internet for a random fact and dynamically builds charts. Takeaway - If you can design it, you can build it. Endless possibilities. I want to keep building this one - what would you add? It's open source, let's build it together.

The internet has run on request-response for 30 years. But Agents broke that pattern in a matter of months. Today agents: - stream - are long-running - push structured/unstructured data at the same time - make tool calls and state updates, all at once and more. If you try putting that behind a REST API, you'll hit wall after wall after wall. That's the core technical problem we solve at @CopilotKit. Instead of forcing agents into the old request-response box, we give them a real-time bridge to your frontend (via the AG-UI protocol) which is becoming the industry standard and has been integrated/adopted by frameworks like @Google ADK, @AWS Strands, @Microsoft Agent Framework, @LangChain, @Mastra, @Pydantic, @crewAIInc and more. Your agent does something? Your UI knows instantly. Your user clicks something? Your agent knows instantly. A live, two-way connection between what your agent is doing and what your user sees. Give your agents true interactivity and generative UI 👇 Start by playing around with our Interactive Dojo: dojo.ag-ui.com/langgraph-type…

Built a template for adding @excalidraw to your AI agents using @CopilotKit — describe any system, get a real diagram back in chat. Edit in local canvas, refine via chat. Auto-saved workspaces to disk, no database. Push live in one click. 👾 Repo: github.com/CopilotKit/exc…

@Anmol_Codes @excalidraw Wow!🪁

Seriously, how cool is this?! If you can design it, you can build it indeed!

CopilotKit 🤝Google 🤝Oracle Build Reusable agents with Generative UIs ✨🩵

Building custom UIs for AI agents usually means rebuilding the same "middle layer" in every stack to keep state and progress in sync. We are changing that 🧵

We worked with @google & @oracle to give portable agents Generative UI. Oracle's Agent Spec now supports Google's A2UI through our AG-UI protocol, which has close, open integrations with both. The AG-UI (Agent-User Interaction) protocol is quickly becoming the industry standard for connecting agentic backends to agentic frontends. This is what happens when you build in the open with interoperability in mind. More collabs like this coming soon 🪁

@Marktechpost You can give your Deep Agents interactive, generative UI with CopilotKit. Here's a starter template demo 👇 github.com/CopilotKit/Cop…

LangChain just released Deep Agents and it points at where agent frameworks are actually heading. Deep Agents adds: • planning with write_todos • filesystem-backed context handling • subagent spawning with task • long-term memory across threads • LangGraph-native execution That means fewer brittle agent demos and a cleaner path to long-horizon workflows. This is what happens when agent design moves beyond “LLM + tool loop.” #AIagents #LangChain #LLMOps #GenAI #AIEngineering @LangChain

@nathan_tarbert @tavilyai Nice, let's go!! 🚀🚀🚀

Happy to share that @AlemTuzlak officially joined us at @CopilotKit 🚀 Alem is the Co-creator of @tan_stack AI & Devtools, and brings super relevant knowledge + experience. A tip from him: "Not sure how to handle agent orchestration? Copilotkit is the key" ...I agree! Looking forward to growing together 🪁🪁