Efficient data sharing from separate systems

Securely access data across databases using chatbot agents with appropriate permissions.
Photo by Sajad Nori on Unsplash
Use Cases
Navigating data silos

The Challenge - access data from cordoned off sources

Organizations often store data on isolated or separate systems, which could be distinct databases, software, or storage systems that operate independently and might not easily share information with each other. This presents a problem when there is a need to share information, while maintaining access restrictions. 

Say, Department A wants to talk to B on the number of person-hours billed during a particular period. Or Department Y needs to know details of items procured for some event. 

Depending on humans for providing access to every department is cumbersome enough, but when the data retrieval has to be done periodically and quickly, it becomes a matter of time and resources. 

The Solution - a chat based model that cites its sources

Separate agents are deployed for each department or silo. They have the same access as the personnel from that department. Agent A can get information from Department A, Agent B from Department B, and so on. 

Agent A for example processes a query on the lines of “How many headphones were purchased in the last 90 days?” and returns the answer along with a reference to the document that it got the information from. As a large-language-model, it can read a document, such as an invoice or a purchase order, go to the section that the query needs answers from, and retrieve the relevant information.

With this solution, we achieve two results:

  1. Controlled interoperability between silos that adheres to existing authorization rules.
  2. An information retrieval technique that uses LLMs, eliminating the need to manually look up the information.

The dataX Advantage

With extensive experience working with machine learning models, including LLMs, our chat bots do not hallucinate. We have deployed B2B product data solutions across industries and domains, and our algorithms can understand product data better than anyone else. Whether it is reading invoices, or blueprints, or going through spreadsheets, dataX gives you the most accurate answers.

More Use Cases