What if integrating your data for AI didn’t take weeks or months – but happened in minutes, without ever moving your data at all?
That’s the vision guiding the development of a new data mapping agent at SAS. It’s not just a feature – it’s a potential shift in how organizations handle one of the most frustrating parts of analytics: data preparation.
“A common task in almost every AI project is to map columns of your existing environments to columns required by AI models,” explains Udo Sglavo, SAS Vice President of Applied AI and Modeling. “What if we could use agentic AI to assist with this tedious task?”
Sglavo and his team created a data mapper agent that uses large language models to automate this process as much as possible.
Data prep is still the elephant in the room
“We all wish data mapping was as simple as a line item in a project plan … but the real complexity behind customer data is staggering,” said John Boyd, SAS Vice President of Solutions Product Management.
He’s not wrong. Before any model can run, teams often spend weeks building custom pipelines – extracting, transforming and loading data across environments to make it usable. And it’s not always smooth sailing.
What if you didn’t have to move the data?
A new data mapper agent is forthcoming from SAS that aims to eliminate all that complexity. It uses automatic schema mapping and virtual views into a customer’s environment, allowing models to work directly off existing data – no duplication, no restructuring and no lengthy setup.
“Imagine a day where integrating systems is no longer a concern,” Boyd said. “Where we show up to a prospect site, ask them to enter a handful of credential information – and show them SAS' complete model portfolio working off their data.”
That’s not a distant dream. The development team has already been testing the data mapper agent on use cases like medical adherence risk modeling, showing how this kind of real-time integration can streamline deployment and reduce cost.
Less risk. More value.
Data management isn’t just a technical challenge – it’s a trust issue. As Boyd put it:
“Customers feel like they are putting their careers in the hands of others when they are dependent on a large data management project.”
The goal of a data mapper agent is to ease that burden. It’s designed to help users tap into the value of their data faster and with fewer hoops to jump through. The initial focus is adding data mapping capabilities to pre-built models and quickly expanding SAS solution offerings.
Instead of spending months wrangling data, teams could focus on solving problems, deploying models and delivering outcomes.
Register for SAS Innovate streaming sessions to learn more
