Time for another update. These past two weeks have been one of those where the whole team has been heads down pushing a lot of the open fronts for the next release. Let me briefly walk you through some of the highlights to give you a glimpse of what has been released already and what to expect in coming releases!
Claude Desktop Extension and MCP Support
Some users were complaining about having to manually edit Claude Desktop’s config file to integrate Baselight into the app, and we agree. To fix this, we have just released the Baselight Claude Desktop Extension so installing Baselight in Claude is as simple as downloading a file and dragging it to your Claude Desktop app. You can find the docs to install the extension here.
Our MCP server is still in alpha, so not all Baselight users have direct access to, if you want to become part of the alpha and test this Claude extension drop us an email (hello@baselight.ai) and I will add you.
And many of you may be asking yourself “wait a minute! ChatGPT already released support for custom connectors through MCP, does this mean that Baselight users will be able to leverage Baselight in ChatGPT too?” We’re on it!
We are improving a few of the authentication flows required to be able to support connecting Baselight to ChatGPT and Claude out-of-the-box, covering a large portion of the LLM market (we can extend this coverage through MCP SuperAssistant, but let’s leave that out of this discussion for now).
Finally, we just merged the last few PRs for the quota enforcements, the last backend piece that we were missing in order to be able to release the MCP integration more broadly to all Baselight users.
Dashboard Improvements
You probably saw in my last update, that we have just released a dashboard. The high-level feedback has been pretty good, but there were still things that could be improved, and this is what we dedicated some time to the past week.
- Titles look nicer and larger (to make our beloved data wizard, Paulo, happy).
- Numerical blocks have received a rework to allow more customisation.
- There was a performance issue when dashboards load that we are still debugging (but should have improved already). So if you see big dashboards taking a while to load for the first time it is not your computer but our network 🙂
- Being able to add titles to the alternate axis was a highly requested feature which is now available.
- We also improved the experience of adding blocks to large dashboards with the floating add buttons and auto scroll.
Baselight AI (bAI) and EVM datasets
And I wanted to leave you with a bit of a cliffhanger of what we are working on:
- We already kicked off the implementation of Baselight AI (bAI), an AI that will allow you to interact with the Baselight platform through natural language. bAI will allow users to ask questions, create queries and dashboards, explore datasets, publish insights, and more without having to bring your own LLM and all from the Baselight platform. The first few tests that we’ve run look gorgeous (and pretty promising).
- Finally, we are doing a lot of work to make our on-chain data a charm to query so you can easily surface insights that were hard to track before, and so you can create beautiful dashboards out of it. On top of this, there are a lot of backend improvements that we are doing to make our data pipelines more efficient and improve the performance of querying these big datasets (this involves buzzwords like Iceberg, dbt, Spark, but I will leave the nerdy stuff for those of you that are interested in the nitty gritty detail).
And with that! See you in two weeks!
Alfonso de la Rocha – CTO
https://x.com/adlrocha