Autodesk Rapidly Expands Insights in Software Using Amazon SageMaker and Amazon EMR

Founded in 1982, Autodesk provides software products and services for several industries, including architecture, engineering, construction, manufacturing, and media and entertainment. Its AutoCAD software helps architects, engineers, and construction professionals create precise drawings and renderings digitally. Autodesk started developing the My Insights service for AutoCAD in 2020, employing usage data to proactively share time-saving recommendations directly with users to improve the experience and soften the learning curve required to use complex AutoCAD features. Because other parts of its infrastructure were already successfully running on AWS, the company knew that it could build My Insights using AWS services and connect it seamlessly to its existing infrastructure.

From the initial release of My Insights, Autodesk chose to use AWS Batch—a fully managed batch processing service that lets developers, scientists, and engineers easily and efficiently run hundreds of thousands of batch computing and ML jobs—to avoid the complexities of manually provisioning, managing, monitoring, and scaling batch computing jobs. Since then, Autodesk has continued to improve My Insights, adding features powered by artificial intelligence using Amazon SageMaker, such as the Macro Advisor service that identifies repeated sequences of operations and recommends automatable command settings directly in the application user interface while users are working. 

The scale of My Insights involves managing multiple terabytes of data daily, requiring Autodesk to think critically about its approach to the design. Autodesk met with an AWS solutions architect to discuss the technical strategy and best practices for managing that scale of data. “Our collaboration with the AWS team has been great because of the brainstorming sessions we’ve had with the AWS solutions architect and the overall guidance that we’ve gotten from AWS,” says Ashish Arora, engineering manager at Autodesk.

Autodesk has been able to automate some of the suggestions for My Insights by running ML workloads on AWS. To support the data processing aspect of its ML models, Autodesk uses Amazon EMR, a big data cloud solution for running large-scale distributed data processing jobs, interactive SQL queries, and ML applications using open-source analytics frameworks. By implementing ML using Amazon SageMaker and Amazon EMR, Autodesk is generating more insights faster. “We increased the number of insights by a factor of 10 by using ML on AWS,” says Arora. “Now, Macro Advisor keeps generating new types of macros to give users insights as they use the product more.” 

For Autodesk’s users, these improvements to the My Insights service simplify the onboarding process for using AutoCAD and help them work smarter in the software by providing information in the moment about relevant shortcuts and commands. “It’s simple to get started with AutoCAD, but there are many layers to it,” says Kaila Tenedorio-Carty, senior marketing manager for AutoCAD at Autodesk. “By using AWS to generate more valuable insights, My Insights helps the user work smarter with different commands and save time with incremental gains by offering suggestions for using fewer keystrokes.” For example, with one common command, My Insights can suggest a shortcut to apply the command in multiple places at once based on the user’s behavior. These suggestions are driven by ML on AWS and give users on-the-fly support and tips right when they need them.

Another significant benefit of using fully managed AWS services is that Autodesk can focus on development rather than operations, saving the company an estimated 20 percent of work on development operations. Autodesk’s homegrown decision engine uses AWS Batch for running business rules and doing targeting, segmentation, and A/B testing, which off-loads the heavy lifting of tasks like provisioning capacity and managing patches from the company’s development teams. “With the native features available in AWS Batch, like automatic scaling, we can scale as needed and manage My Insights without worrying about the operational tasks across all the different loads that we use,” says Arora. “Using AWS Batch, we can automate tasks so that we can focus on improving our products rather than managing operations.” Additionally, with the scalability of AWS Batch, Autodesk is prepared for the future and can expand its products and features as needed without investing in additional hardware.

Because of the scalability and stability of AWS services, Autodesk is prepared for future expansion. The company plans to continue developing additional features for My Insights for AutoCAD while exploring what’s possible in its other software products. “Using services like AWS Batch and features like automatic scaling, we were able to future-proof our software so that we wouldn’t need to worry about scaling our own servers as we expand,” says Arora.