TLDR: Amazon Web Services (AWS) has announced the public preview of inline code nodes within Amazon Bedrock Flows. This new feature allows developers to directly embed custom JavaScript and Python code snippets into their generative AI workflows, enabling more precise data manipulation, formatting, and seamless integration with external APIs. The enhancement aims to bridge the gap between low-code simplicity and programmatic flexibility, streamlining the development and deployment of sophisticated AI applications by reducing the reliance on separate AWS Lambda functions.
Amazon Web Services (AWS) has unveiled a significant advancement in its generative AI toolkit with the public preview of inline code nodes in Amazon Bedrock Flows, announced on August 21, 2025. This strategic enhancement is designed to empower developers by allowing them to embed custom JavaScript and Python scripts directly into their AI workflows, thereby offering unparalleled flexibility and precision in data handling and external system integration. The move addresses a growing demand from enterprises for hybrid AI development approaches that combine the ease of visual, low-code builders with the power of custom programming.
The primary purpose of these inline code nodes is to facilitate advanced data manipulations and integrations that extend beyond Bedrock Flows’ native capabilities. Developers can now perform tasks such as string parsing, complex data formatting, and direct integration with external APIs without needing to exit the visual flow builder or set up separate AWS Lambda functions. This significantly streamlines preprocessing and postprocessing tasks, such as normalizing input data before it reaches a large language model (LLM) or formatting model outputs directly within the flow. The feature supports Python 3.12 and above, handling code with a binary size of up to 4 MB, and also supports popular packages like opencv, scipy, and pypdf. For JavaScript, it supports up to 4KB of code per node.
Execution of these inline code nodes occurs within a secure, AWS-managed, sandboxed environment, ensuring that custom logic operates in isolation and without internet access, leveraging AWS Lambda under the hood for scalability. Users can define inputs and outputs using JSON schemas, ensuring seamless integration with other nodes, such as prompt or agent invocations. This capability is particularly beneficial for sectors like finance and healthcare, where precise data transformation and integration with existing systems are critical.
This introduction builds upon previous updates to Amazon Bedrock Flows, including the persistent execution for long-running workflows, which was announced in a June 2025 update. The combination of long-running support and inline code enables workflows to handle asynchronous tasks, such as processing large datasets or waiting for external approvals, while embedding code for real-time adjustments. Industry experts note that this integration reduces the need for separate Lambda functions or external services, potentially cutting development time and costs, and accelerating enterprise adoption of generative AI solutions by simplifying development and reducing maintenance overhead.
Also Read:
- AWS Bolsters AI Agents with Predictive Machine Learning Integration and Model Context Protocol
- AWS and Workato Forge Alliance to Scale AI-Powered Workflow Automation
Security and traceability remain paramount, with Bedrock Flows incorporating built-in guardrails and logging for all code executions. The inline code node is currently available in public preview in select AWS regions, including US East (N. Virginia), US West (Oregon), and Europe (Frankfurt), inviting developers to test and provide feedback for iterative improvements.
