Qumulo has built a Microsoft Copilot AI connector 

Azure Native Qumulo now integrates with Microsoft Copilot and Graph connectors. The aim is to access and analyze unstructured data, delivering detailed insights based on natural language prompts. 

Sean Gwaltney, who looks after Azure Development & Strategic Engagements at Qumulo, made a LinkedIn post last month saying: “Quietly, without making big promises the Azure Native Qumulo team put their heads down and built a deceptively simple solution to connect your unstructured data to MicrosoftCopilot in a functional manner.”

What scaleout file storage supplier Qumulo has done is to use Microsoft Graph and its customized connector facility to make files stored in its Azure instantiation, Azure Native Qumulo (ANQ), available to Copilot. It can then respond to users’ conversational inputs by referring to ANQ-stored information. This is retrieval-augmented generation (RAG) in action. This is accomplished by having a Qumulo-provided connector feeding the files to Graph and having its semantic index facility vectorize them.

A Microsoft web doc says: “A vector is a numerical representation of a word, image pixel, or other data point. The vector is arranged or mapped with close numbers placed in proximity to one another to represent similarity. Unlike a standard keyword index, vectors are stored in multi-dimensional spaces where semantically similar data points are clustered together in the vector space, enabling Microsoft 365 to handle a broader set of search queries beyond “exact match.”

Qumulo Microsoft Copilot video

A Qumulo video talks about this. It demonstrates a financial services-oriented environment with millions of invoices, using the Microsoft Copilot interface to query for details on specific invoices and customers. Customised connectors can enable access and insights for multiple file types stored in the Azure Native Qumulo environment. 

The video commentary says: ”if you’re a financial services institution you probably have digital records everywhere on-prem, in the cloud and in archives, and in every type of format; databases, PDF files, office documents, even text files. 

“We’re talking literally millions of files, maybe even billions, with different formats and different layouts in all kinds of different places.

“You want to be able to search and analyze all those files but most legacy tools only work with structured data, and AI engines like ChatGPT can pose both security and competitive leakage risks.”

Once the custom Qumulo “connector has been activated, millions of PDF files can be read via Copilot’s Ai and used to answer business queries from within Microsoft 365. … Each connector can be fully customized so you can import and use whatever file data you need, whether you’re consolidating records after an acquisition, searching old data for legal discovery, or itemizing and tracking archived records.  … you can also use custom connectors to import data from different file types or from different locations.”

There is no leakage of proprietary information outside the customer’s Azure environment, we’re told, and you don’t need to be a SQL wizard or wannabee data scientist to get answers to quite complex questions.

Comment

This work by Qumulo supports Nasuni’s contention that Microsoft CoPilot has already won the race to provide GenAI chatbot facilities inside Microsoft’s Azure Windows and 365 environments. It seems inevitable that all file storage suppliers for Windows and Microsoft 365 will do the same, aiding their customer’s RAG efforts. Will on-premises Windows server systems do the same thing? We don’t see that happening as CoPilot is an Azure cloud-only feature. This could facilitate a renewal of interest in having unstructured data migrated to Azure.

As chatbots become available on premises (see Nutanix’ GPT-in-a-Box AI initiatives), then making a file – and object – storage supplier’s customer-stored data available to the chatbot for RAG will become necessary.