@trq212 Great examples and I’ll try them out. I’ve been exporting HTML for a while. It’s great to have in team meetings to show interactive examples on complex topics.
Just built a Python-based MCP server that connects Claude Desktop to the Autodesk Platform Services API.
Natural language queries → Real API calls.
Here’s a quick intro: 🎥👇
@_5upr4@_philschmid@GoogleDeepMind OpenAI provides a way to do this in their AgentSDK. I’m assuming it might be a similar method. I haven’t tried with Gemini yet tho.
https://t.co/1ywcy0CuUi
JFK Files in Text
I used Replit AI on my phone to download the files, use Gemini Flash to OCR the pdf to convert to text files, and put it up on GitHub.
You can search directly on GitHub or download or use an LLM to analyze. Link in thread.
@morganb@vercel@Replit I’ve been thinking about this, I hear a lot of ideas and requests and this is a great way to triage those better. Requiring a POC with the idea.
Exploring new ways to interact with Autodesk data.
Inspired by a conversation at #AU2024, I built a POC connecting an iOS app, OpenAI Realtime API, LiveKit, and Autodesk APIs for real-time data access.
Demo video attached.
#Autodesk#OpenAI#LiveKit#iOS#ConstructionTech
@SullyOmarr@ottogrid_ai Such a good question! Maybe these kinds of solutions need outcome-based pricing. If you can prove it cuts a company’s research costs, then it’s worth a piece of those savings.
Paul Graham wrote an essay on this.
In summary, prioritize building strong connections with early users, delivering exceptional service, and refining your product. It’s crucial for learning what users truly want, building loyalty, and establishing a strong foundation for growth