🤖 ᴺᴱᵂ ᴾᴸᵁᴳᴵᴺ Azure AI - ChatGPT Streaming on your Data inc. Function Calling & Markdown [Keeps your keys secure]

Hey Bubblers!

Harness the power of Azure AI to unlock your data’s potential with lightning-fast searches and context-aware responses, powered by the innovative capabilities of OpenAI’s ChatGPT models.

Import data in formats like txt, md, html, Word files, PowerPoint, PDF or even websites.
Our platform ingests and chunks data into Azure AI Search, providing a seamless analysis and conversation experience.

Lastly, Azure Blob storage container integration offers easy data access, empowering analysis and conversation using Azure OpenAI Service. Whether data is structured or unstructured, our platform adapts, ensuring confident data-driven decisions.

You can test out our Azure AI - ChatGPT Streaming on your Data Plugin with the live demo.

Microsoft-Azure-OpenAI-service-is-now-available

Enjoy !
Made with :black_heart: by wise:able
Discover our other Artificial Intelligence-based Plugins .

4 Likes

Great job. Bookmarked! :+1:

1 Like

I am using the pluging. But when I try to save the messages using the All Current Threads (Raw Data), the latest response is always missing.

I need to send another prompt so that the second to last response would be saved to my database. Could you help?

Also, is there a way to set how many messages it will “look back”? Or is there a way to make it more affordable to use the thread? If not, can you add an action for “Completion”?

Thanks, in advance for your help.

grady

Contacting you in DM @gradylam

Yes, there was a bug that has been fixed @gradylam .

As discussed, delete threads feature has been added to account for this.
Embeddings may come at a later stage for extended context.

Hey Bubblers!

Informing you that I have added functions to this plugin, enabling Bubble workflows to be triggered on user prompt and feed data back to the LLM.

Enjoy!

Hi! Does this use a separate server for message response streaming back from Azure? Or are you handling this directly in Bubble?

Also, how are you chunking the data for Azure AI Search? That is, how are you determining the size of each chunk, etc?

Yes, it uses a separate infrastructure to keep your keys hidden from the browser and provide streaming.

If you use Azure AI Studio, it is automatically done on their side upon data ingestion, otherwise manual but in both case it is independent from the plugin features.