Tinybird

Claude made this dashboard. All we had to do was ask. We just wrapped up a natural language chat in which we learned a ton about Bluesky 🦋. Our chat was powered by the #Bluesky Firehose API, Anthropic's Model Context Protocol (#MCP), #Claude, and Tinybird. Here's what we learned about the growing distributed social media platform from just a few hours of data: - About 20k new follows happen every minute - About ~1.15M unique users have posted content - Approx 7.7M posts since we started tracking - The average user has made ~7 posts - 2.1% of users have posted 50+ times - 15.7% of users have posted 10-49 times - 82.2% of users have posted <10 times That's just the surface of what we could explore. And it begs the question: What can you ask of your data now that MCP makes LLM applications more readily aware of specific contexts? Want to read the full transcript of our chat? It's published as our latest blog post, co-authored by Claude himself (itself?) and Tinybird's MCP Server 😏 Read: tbrd.co/claude

It's super cool to work in a company where it takes us 1-2 days to have something in production since Anthropic released it 🚀 Basically saying "C; C; A; so cool, could you build a dashboard" you have it https://lnkd.in/esMjyFGK

Imagine simply chatting with Claude to explore and work with the data you are capturing in real-time. Let it generate the queries and push them down to Tinybird to execute, get a summary of the findings and produce some charts! Pretty cool stuff.

Why build yet another chat interface when Anthropic already built Claude? Following MCP's release, we've built v0 of the Tinybird MCP Server so you can chat with any data in Tinybird straight from Claude Desktop. It's early days, and I'm not entirely sure where this is headed and what it might look like days/weeks/months from now...but we can already do some pretty cool things. How about analysing billions of rows of fresh, hot BlueSky data in real time without writing any SQL? Using Bluebird https://lnkd.in/ebu6zBr5 I can push the real time event stream from BlueSky to Tinybird. Using the Tinybird MCP Server https://lnkd.in/eBA8dbfY I can connect Claude Desktop to Tinybird. And then I can just...chat with Claude to work with my data. Claude is generating the queries, pushing them down to Tinybird to execute, giving me a summary of the findings and producing charts! Pretty crazy stuff.

Tinybird MCP Server v0 ✅

Yesterday, Anthropic announced their open-source Model Context Protocol (MCP), which is intended to provide a universal, open standard for connecting LLM applications to local and remote data sources to improve context awareness for LLM applications and assistants. We're already thinking about MCP at Tinybird, and how we can use it to solve some interesting problems: - Give your AI assistants context about your specific Tinybird Workspaces, the resources you're working with, use cases you're building, metadata about your data sources, columns, endpoints, etc. - Giving you the ability to quickly generate Tinybird-focused MCP servers so that LLM applications can utilize your published Tinybird APIs for context, giving your customers access to resources published via Tinybird to assist in their own development or use cases. LLMs are not just changing the way we code, they're changing the way we build teams, startups, and product roadmaps. It's amazing what a small handful of developers can create - and quickly - alongside AI. We're continuing to look at ways we can make the Tinybird experience more aligned with the ways you work - including how you use LLMs and AI assistants. As we evaluate MCPs and other AI initiatives, what would you like to see from Tinybird?

🎉 Join Us at re:Invent! 🎉 🌟 Stop by Booth #1856 for exclusive giveaways and demos! ✨ Meet our team and discover how you can create real-time data apps on AWS with Tinybird. #AWSreInvent #AWS_Partners

🚀 "Because of Tinybird's capabilities, we've changed our roadmap, because what's possible with Tinybird just wasn't possible with anything else." - Carl-Einar Thorner With Tinybird, Zembula was able to transform their email personalization platform from timing out on basic queries to delivering sub-second responses on 20+ billion messages unlocking real-time creative analytics at scale. Want to know how they accomplished this in just 4 weeks? Full case study ➡️ https://tbrd.co/zembula

Speed wins. That means being able to quickly do things yourself. Things that get in the way? - Struggling to find examples to help you get started. - Being confused about how to use a feature. - Having to wait for a response to a troubleshooting question. All things that can be prevented with great docs. Now you can help us continuously improve our docs with the new docs feedback widget.