AI software development shifts as AWS and poolside team up

Amazon Web Services and poolside have announced a multi-year agreement to make poolside's generative AI Assistant and foundation models (FMs) available in Amazon Bedrock. Amazon Bedrock is a fully-managed service that offers developers access to high-performing models from leading AI companies through a single API. As a result, enterprise customers will soon be able to customize poolside's generative AI Assistant for software development with their own data, leveraging the security, privacy, and performance of AWS.

AI software development shifts as AWS and poolside team up

As part of the agreement, poolside will also leverage AWS Trainium chips to power inference for its malibu and point foundation models, enabling optimal price performance for customers. AWS Trainium is a high-performance machine learning chip designed to reduce the time and cost of running generative AI models. AWS customers are now able to deploy poolside securely on Amazon Elastic Cloud Compute (EC2) and will soon be able to so through Amazon Bedrock, relying on Amazon EC2 Trainium instances for inference.

Solving for the generative AI needs of enterprise organizations

Bringing poolside's models to Amazon Bedrock and EC2 provides enterprise customers of both companies with a unique advantage to solving complex software engineering tasks and driving development productivity for their organizations through generative AI. poolside FMs are designed to be fine-tuned with each business's code and data, producing a proprietary generative AI model and software engineering assistant for that specific business. poolside in Amazon Bedrock and EC2 will enable enterprise organizations to meet security, privacy and compliance concerns and quickly deploy their custom generative AI models. AWS customers will be able to deploy poolside models where their data is stored, within their firewalls, on the world's most secure cloud, and with no data shared back with poolside.

poolside's Reinforcement Learning From Code Execution Feedback (RLCEF) technique runs on AWS today, powering approximately one million container images, while allowing for 10,000 code executions per minute. RLCEF allows poolside to generate synthetic training data at scale by exposing models to realistic coding tasks in real-world code bases, providing feedback based on code execution, and employing a large-scale Reinforcement Learning approach for training. Proving RLCEF at scale is what allows poolside models to train on increasingly more data without relying on any of their customers' data – enabling enterprise customers to deploy models securely and privately.

"We've been incredibly impressed with the AWS team and are excited to partner on this journey to unlock more of the potential of AI for software development. Today, the majority of developers sit within an enterprise environment where access to these tools is scarce and their full potential is not yet realized. Companies of this scale need a tailored model that can both capture their proprietary knowledge and learn from their interactions over time. That is what we've built, and AWS's reputation and depth within the enterprise is key to accelerating adoption and impact," said poolside CEO Jason Warner.

"Giving AWS customers the flexibility to run poolside in Amazon Bedrock or on their own EC2 instances will further enhance their ability to innovate with generative AI, poolside offers a unique solution for software engineering teams of large enterprise customers looking to build and scale generative AI applications through foundation models without sacrificing privacy and security. It's yet another way for enterprise organizations to leverage their internal knowledge base to build proprietary, tailored solutions," said Dave Brown, Vice President, AWS Compute & Networking Services.

Streamlined procurement, contracting, and budgeting through their relationship with AWS makes it even easier for AWS customers to experiment securely with poolside.

Models for specific use cases

AWS is the first cloud provider to offer fully-managed models from poolside. The poolside Assistant is powered by models ideal for software development and coding-related use cases: poolside's malibu and point generative AI models. poolside's FMs are trained on the world's largest pool of code execution data and fine-tuned on how large enterprises write software. malibu is highly versatile and excels at tackling complex engineering challenges, delivering intricate code generation and test and documentation generation. point is engineered for rapid code completion, leveraging advanced context awareness to accurately predict developers' needs. point offers real-time suggestions for algorithm optimization and design patterns. With poolside, Amazon Bedrock customers will soon have their own proprietary models, fine-tuned on their own data, that continuously learn how their developers write code.