Microsoft Business Central deploys advanced semantic search in six months with Azure SQL Database

Data is at the heart of every successful business. The Microsoft Dynamics 365 Business Central team is harnessing native vector capabilities from Microsoft Azure solution Azure SQL Database to empower customers with natural language queries that help them find the right data, at the right time.

Simplifying data searches for more efficient operations

Continuous improvement is a core tenet for the team behind Business Central. For 50,000 small and medium-sized businesses, the Business Central enterprise resource planning (ERP) solution powers business operations ranging from finance to sales and supply chain management. ERPs often use specific terminology to categorize information, and Microsoft wanted to make it easier for customers to navigate Business Central using their own words.

Customers also needed to be able to find information based on general product descriptions. “Think of an email stating that a customer wants to buy a blue chair that you have in the store, but they do not know the item number or name,” says Jens Møller-Pedersen, Software Engineering Manager at Microsoft. “We wanted to add searching by meaning to find items with just a description.”

The team needed to support these semantic searches in the complex, multi-tenant Business Central environment. One option would be to incorporate a standalone search service with indexing, but the Microsoft team opted for a bold alternative—using its existing Azure SQL Database infrastructure to combine relational and semantic searching to respond to natural language queries.

Deploying transformative search capabilities in just six months

To drive this innovative solution, the Business Central team became an early adopter of the Azure SQL team’s full vector support offering, so they didn’t need additional infrastructure. “Each new service that we add to our architecture requires a lot of work across different teams,” says Agnieszka Potulska, Senior Software Engineer at Microsoft. “Since we already had native vector support with Azure SQL Database, we only had to write the code that will search in the vector space.”

Business Central’s multi-tenant environment uses elastic pools in Azure SQL Database to scale seamlessly with a growing customer base. “We know that Azure SQL Database easily scales to 50,000 databases, half a million databases, whatever we need,” says Møller-Pedersen.

Within six months, the Business Central team had embedded vectors that would power three new features. Tell Me and Chat with Copilot help people find reports and data with semantic searches, and Dynamic Tool Mode connects AI agents with data workflows. These transformative search capabilities were developed and deployed with little operational cost. 

Powering natural language queries for 50,000 global customers

These new features are available for all of Business Central’s 50,000 customers, offering a modernized experience that helps lower the entry barrier for new users. “You can interact with Tell Me or Chat with Copilot in your own language using your own words. For example, you don’t need to know whether something is called a sales quote or an offer,” says Møller-Pedersen. “Navigating the product has become significantly easier.”

The close collaboration between the Business Central and Azure SQL teams showcases how Microsoft is developing novel approaches to improve the customer experience. By incorporating an Azure SQL Database solution, the Business Central team introduced new capabilities within its existing infrastructure. “You can modernize how you interact with your data without changing how you store it,” says Møller-Pedersen. “You can look at your existing data in a completely different way.”

Accelerating innovation to streamline business operations

The Business Central team is eager to build on its current success by continuing to expand its semantic search capabilities. The team is working closely with the Azure SQL team on using DiskANN indexing, which will help customers and AI agents efficiently use natural language search in the vast amount of relational data in Business Central.

For the Business Central team, this experience also illustrated that the most impactful changes are not always the most complicated ones. “I think our biggest success is how seamlessly we’ve been able to put this into production,” says Møller-Pedersen. “I don’t recall any operational problems rolling out this new functionality to our customers.”

Discover more about Microsoft on Facebook, Instagram, LinkedIn, X/Twitter, and YouTube.