AWS re:Invent 2025 - Automatic Enterprise ML with Amazon SageMaker and Nextworld (SMB201)

Automatic Enterprise ML with Amazon SageMaker and Nextworld

Overview of Nextworld

• Nextworld is a business applications platform company that has created a no-code platform powered by AI, leveraging Amazon SageMaker.
• The platform is designed to enable non-technical users to build and deploy AI-powered business applications.
• Key features of the Nextworld platform:
  • AI-native: Uses AI to help build applications, with assistants and agents that can create apps from natural language.
  • Bakes AI into all applications, exposing them via APIs and chatbots.
  • Enterprise-grade with audited, managed infrastructure.
  • Built on an ERP (enterprise resource planning) heritage, handling complex business applications.

Nextworld's Approach to Machine Learning and AutoML

• Nextworld previously had a traditional machine learning facility, but it was geared towards experienced data scientists and had a lengthy development cycle.
• They explored building their own AutoML solution, but ultimately decided to leverage Amazon SageMaker Autopilot.

Benefits of SageMaker Autopilot

• Automated the entire machine learning lifecycle, handling feature engineering, algorithm selection, and model tuning.
• Enabled Nextworld's product owners, not just data scientists, to build accurate ML models.
• Provided a serverless, API-accessible solution that fit Nextworld's multi-tenant architecture.
• Offered advanced model analytics and visualization capabilities.

Real-World Results

• Nextworld saw a 95% improvement in model build time, reducing it from 4 weeks to 2 days on average.
• Achieved a 33% average improvement in model accuracy compared to manually built models.
• One customer saw a 25% accuracy improvement, going from 75% to 90% accuracy, in a life-critical organ viability prediction use case.
• After releasing AutoML, 100% of Nextworld's customers up for renewal not only renewed but also upsold, resulting in a 54% average contract value increase.

Technical Architecture

• Nextworld's AutoML solution leverages SageMaker Autopilot's serverless endpoints for training and inference.
• Users interact with the AutoML functionality through Nextworld's browser-based client, which communicates with Aurora RDS for data and the SageMaker Autopilot APIs.
• The serverless architecture provides cost savings and scalability for Nextworld's customers.

Conclusion

• Nextworld has seen significant value from integrating SageMaker Autopilot into their platform, both for their customers and their own business.
• AutoML has enabled non-technical users to build accurate, production-ready ML models, freeing up Nextworld's data scientists to focus on platform development.
• The combination of Nextworld's no-code AI platform and SageMaker Autopilot has proven to be a powerful solution for enterprises looking to leverage machine learning without the traditional barriers.