
Building @CopilotKit. Functional programming, generative user interfaces, and software design. @Angular GDE. He/him/y'all.
Portland, OR
Joined June 2008
- Tweets3.2K
- Following533
- Followers6.6K
- Likes7.5K
My team at @CopilotKit launched OpenBot on Product Hunt today.
I got to do some deep technical work on streaming threads for this product, built with Elixir and Phoenix.
An upvote and a comment would mean a lot! producthunt.com/products/cop…
Mike Ryan retweeted
Been working on Graft. OSS, agent-first tooling for maintaining forks and patches while getting upstream fixes automatically using agents. Works with npm, Rust, Go, and Python, package managers, etc.
github.com/brandonroberts/gr…
Mike Ryan retweeted
It’s wild how well Generative UI works with A2UI 1.0, Express & macros 🤯
A few lines → a full dashboard.
@GeminiApp Flash Lite in the cloud.
Gemma 4 for self-hosting.
Powered by @mastra & @CopilotKit.
Watch it in action 🎥
Live demo in the comments 👇
🚀 We launched OpenBot in August - beyond excited to have reached the 1.0 today
OpenBot was re-designed and renamed in the days following the release of agent workspaces by the big labs.
We saw a strong need for an agent workspace that is open, self-hostable, customizable, and not tied to big token.
@CopilotKit has been leading the agent-user interaction layer for years, and our building blocks are trusted by the majority of the Fortune 500.
So we knew we could ship an amazing product, that can plug in to any agent harness (via AG-UI), and can be enterprise-grade from the start.
The launch got over 850k views on X, and much more elsewhere and OpenBot quickly became the leading open agent workspace.
Our releases of OpenDots, OpenMuse and OpenDuet received millions of views 16k more GitHub stars.
Today - these are all rolled into OpenBot as UX 'skins' with important milestones in the 1.0 release. Including:
🖥️ Desktop app, runs fully local
🧠 Any model, or your existing Claude/ChatGPT/Gemini/xAI plan
🔌 Bring any agent: Claude SDK, LangChain, ADK, Strands (AG-UI)
🎓 Teach a Bot: record a task once, it learns it
🤖 Subagents that ping you when needed
🔮 Generative UI: charts, tables, cards
🧰 Hundreds of apps via Composio and Arcade
🏢 Enterprise: SSO, SCIM, audit trail, self-host
The goal is simple:
A great experience for end users.
Real governance for enterprises.
Any custom agent for developers.
Open, and built for all three.
Try it at open.bot
🎉 Introducing 𝙾𝚙𝚎𝚗 𝙱𝚘𝚝
An open source Grok Bot that works with ANY agent harness, designed for real companies.
It includes:
- AI Coworkers
- Generative UI
- Computer use (remote/local)
- Agent-human handoffs
- Full data recording, owned by you
Repo → github.com/CopilotKit/openbo…
We're using this internally at @CopilotKit and it's changing the way we work forever.
Powered by CopilotKit and AG-UI.
More info below 👇
Mike Ryan retweeted
Introducing OpenBot 1.0 🚀
The customizable, open source AI coworker platform your company can actually own and self-host.
It has everything you'd expect and want:
1. Great UX
2. Computer-use
3. Web search and integrations
4. Routines
5. Security
AND so much more, available only with us:
⚙️ Works on your infra + stack
🔌 Plug in any agent (via AG-UI)
👔 Enterprise-ready
🧠 Continual Learning
🔮 Generative UI
1.0 brings everything we've worked on into one comprehensive product, now available on Mac and Windows!
Try OpenBot now: open.bot
Need this for business? Talk to us and we'll support you from day-0.
Follow @0penBot for more.
Mike Ryan retweeted
Cool thing about CopilotKit is that you can use the agent store just like a regular signal store.
Both the agent (via AG-UI) and your code are modifying the same state and the UI updates in real time
![TypeScript code in trip-planner.ts. An Angular component app-trip-planner has a template with a "Plan my trip" button that calls plan(), and an ordered list that renders each stop with @for (stop of trip().stops; track stop). In the class TripPlanner, highlighted: readonly store = injectAgentStore('planner'), and readonly trip = computed(() => this.store().state() as { stops: string[] }). The plan() method gets agent from this.store(), calls agent.setState({ city: 'Lisbon', stops: [] }) (highlighted) and then agent.runAgent().](https://pbs.twimg.com/media/HWNKXmKWgAAf8Yw.jpg?name=small&format=webp)
ALT TypeScript code in trip-planner.ts. An Angular component app-trip-planner has a template with a "Plan my trip" button that calls plan(), and an ordered list that renders each stop with @for (stop of trip().stops; track stop). In the class TripPlanner, highlighted: readonly store = injectAgentStore('planner'), and readonly trip = computed(() => this.store().state() as { stops: string[] }). The plan() method gets agent from this.store(), calls agent.setState({ city: 'Lisbon', stops: [] }) (highlighted) and then agent.runAgent().
Mike Ryan retweeted
CopilotKit fully supports Vue 💚
Everything the React SDK does, now native to Vue:
- CopilotChat, Sidebar and Popup
- Frontend tools and shared state
- Human in the loop
- Generative UI with your own Vue components
Same agent, same AG-UI events. Just a different frontend.
This was a massive contribution by @enekes_abel at @fynk_com 👇
@cwintoniak
Genuinely the biggest criticism of this pattern was the boilerplate, and to some degree tracing the code.
Agents don’t care. Pick a state management pattern that optimizes for correctness.
Maybe Redux, or a Redux-like framework needs a comeback
Having state in ONLY ONE PLACE sounds pretty beneficial to agents
I.e. restrict the decision space, only allow one way of doing things
1
5
1,064
Mike Ryan retweeted
Maybe Redux, or a Redux-like framework needs a comeback
Having state in ONLY ONE PLACE sounds pretty beneficial to agents
I.e. restrict the decision space, only allow one way of doing things
Ready to bring agentic AI into your Angular dashboard? 🤖🛠️
With CopilotKit integration, you can equip LLMs with frontend actions and app context so users can navigate and edit data with natural language.
What’s the first AI feature your users requested? 👇
Mike Ryan retweeted
🚀 Introducing CopilotKit Soar
AI coworkers your company actually owns.
We open sourced the big lab AI products - Grok Bot, Muse, Dots...
Now we're turning their best aspects into one business-ready version that's:
- Self-hostable
- Fully customizable
- Works with any agent harness / model
- Has continual learning
- Available on Web, Mobile, Channels
and more via the AG-UI Protocol, OpenBot and @CopilotKit Intelligence.
By getting early access to Soar, our FDEs will personally support your company in setting up and maintaining the product with your own branding and infrastructure.
Get early access: go.copilotkit.ai/soar-access
(to skip the line, schedule a meeting)
Stop renting your AI and start owning it.
Mike Ryan retweeted
When I was getting into programming the popular thing to do and show of your skills was to create a React TODO app or a movie app.
Today it's migrating typescript to a different language.
Mike Ryan retweeted
Learn how to build Agentic UIs with Angular! 🚀
Join our free live-coding webinar on October 23, where we'll build an Agentic Analytics App using Angular, CopilotKit, AG-UI & A2UI.
Register here:
ai-poland.pl/free-workshops/…
It's straightforward to build an agent inside your product that works well in a demo, but making it actually useful requires a deep understanding of how your users get work done in your app.
CLiF lets CopilotKit Intelligence learn from user behavior outside of chat. It understands workflows and then generates new Skills you can feed into your agent.
INTRODUCING: CopilotKit CLiF
A platform for Continual Learning from Full-stack Trajectories.
Teach your agents how work gets done by turning the work users & employees do into:
→ Exportable datasets (of trajectories)
→ Evals (bring into platform of choice)
→ Skills
→ Fine-tuned models
Your users already do the exact tasks you're trying to teach your agent, every day.
Yet months of evals and engineering leave out your most important dataset...
Product trajectories.
AG-UI already sees every step your agent takes. Now it can also securely capture what your users do: what was on their screen, the buttons they clicked, the server requests they triggered.
Together that's a full-stack trajectory.
Your people already do the work.
Now agents can learn it from them.
Compatible with your existing observability and evals stack. Fully self-hostable.
@CopilotKit CLiF is in private alpha.
Book a demo / Get early access: go.copilotkit.ai/get-clif
More below 👇
Mike Ryan retweeted
INTRODUCING: CopilotKit CLiF
A platform for Continual Learning from Full-stack Trajectories.
Teach your agents how work gets done by turning the work users & employees do into:
→ Exportable datasets (of trajectories)
→ Evals (bring into platform of choice)
→ Skills
→ Fine-tuned models
Your users already do the exact tasks you're trying to teach your agent, every day.
Yet months of evals and engineering leave out your most important dataset...
Product trajectories.
AG-UI already sees every step your agent takes. Now it can also securely capture what your users do: what was on their screen, the buttons they clicked, the server requests they triggered.
Together that's a full-stack trajectory.
Your people already do the work.
Now agents can learn it from them.
Compatible with your existing observability and evals stack. Fully self-hostable.
@CopilotKit CLiF is in private alpha.
Book a demo / Get early access: go.copilotkit.ai/get-clif
More below 👇
Mike Ryan retweeted
🎉 Introducing 𝙾𝚙𝚎𝚗𝙳𝚘𝚝𝚜
Self-hostable, always-on AI coworkers that works with ANY agent harness.
Includes:
- Computer use: browser, terminal & files
- Bring agents to Slack, Teams etc
- Spaces and Pages for projects
- Voice calls
- Web and Mobile
Repo → github.com/CopilotKit/OpenDo…
Powered by @CopilotKit and AG-UI.
Clone this template and customize it however you want. Enterprise-ready.
Mike Ryan retweeted
Introducing AG-UI 1.0
The open protocol for how ANY agent communicates with ANY application.
Now with a stable, versioned contract.
Features included in 1.0:
1. Metadata - send backend data to the UI
2. Subagents - expose delegated agent work
3. Interrupts + resumption
3. Run outcomes and token usage
4. Forward compatibility
Capabilities that have been part of AG-UI (streaming, tool calls, shared state, reasoning etc) are still covered.
Thank you to those who shaped 1.0 🪁
Including builders at @Anthropic, @Google, @Microsoft, @Pydantic, @TanStack and more.
Their feedback led directly to several critical improvements in the final release.
AG-UI is built with those who build on it.
👉 Read more: go.copilotkit.ai/AG-UI-v1.0
The 1.0 spec will remain unchanged, while future protocol evolution will happen through new draft specifications and open community discussion, led by @CopilotKit.
More details below
Congrats to @HeyGarrison and team on the launch of @computesdk Actions. Excited to build with it. 💪
🏁 ComputeSDK Actions are live
What we built:
⚡ Massive concurrency for parallel dev cycles
🌐 Sandbox providers + regions
🌿 Git-first workflow portability
🛡️ Ordered fallback for reliability
🤖 Agent-native CLI + API
Let’s get into it ↓
1
5
1,337
Mike Ryan retweeted
How do you write product requirements for an AI agent?
“Add an assistant” leaves a lot unanswered. When does it ask permission? How can users change their minds?
AG-UI gives PMs a concrete vocabulary for designing how people and agents work together.
Mike Ryan retweeted
We released our AG-UI DevTools Chrome extension.
If you are already working on Agentic UI with AG-UI (via @CopilotKit), this makes your life much easier. If not, now's the best time to try it 😁
Kudos to @wolfmanfx (@SoveriusAI) for the implementation.
chromewebstore.google.com/de…

















