Amazon Bedrock Flows now generally available with enhanced security and traceability

Today, we are excited to announce the general availability of Amazon Bedrock Flows (formerly known as Prompt Flows). Bedrock Flows lets you quickly build and run complex generative AI workflows without writing any code. The main benefits are:

1. An intuitive visual interface simplifies generative AI workflow development.
2. Seamless integration of modern foundation models (FM), prompts, agents, knowledge bases, guardrails, and other AWS services.
3. Flexibility to define workflows based on business logic.
4. Reduce the time and effort it takes to test and deploy AI workflows using SDK APIs and serverless infrastructure.

Bedrock Flows makes it easy for developers and businesses to harness the power of generative AI, enabling them to create more sophisticated and efficient AI-driven solutions for their customers.

Thomson Reuters transforms the way professionals work by delivering innovative technology and GenAI, powered by trusted expertise and industry-leading insights.

“The mission of the Thomson Reuters Enterprise AI Platform is to help our subject matter experts, engineers, and AI researchers bring trusted, cutting-edge technology into the hands of our customers and collaboratively develop Gen-AI capabilities that will shape the future. Amazon Bedrock Flows lets you create complex, flexible, and serverless flows that are easy to evaluate, compare, and version. SDK API for execution You can also use Bedrock Flows to quickly integrate flows with your applications without wasting time on deployment or infrastructure management. I’m excited about the improvement and acceleration.”

– Laura Skylaki, Vice President, Artificial Intelligence, Business Intelligence and Data Platforms, Thomson Reuters.

Dentsu Creative is a global creative agency network dedicated to creating meaningful connections between brands and consumers.

“We leveraged Amazon Bedrock Flows to transform our customer experience. We used Bedrock Flows to accelerate the process of reshaping our books into a format that is more readable for readers with learning disabilities. Bedrock Flows also makes it easy to connect your customer service solution to underlying models like Claude Haiku to respond to common inquiries, saving time and freeing your customer support team to focus on more complex requests. Now you can. Bedrock Flows provides transparency and visibility into generative AI solutions within organizations by enabling non-technical users to understand how AI and business logic are applied using an intuitive visual interface. Whether it’s reaching new audiences or expanding customer requests, Dentsu continues to innovate with cutting-edge generative AI technology powered by Amazon Bedrock Flows.

– Thiago Winkler, Executive Director of Operations, Dentsu Creative Brazil

New features in Amazon Bedrock Flows

Organizations leveraging generative AI need robust safety controls and clear visibility into AI workflows. Today, we are announcing two new features in Amazon Bedrock Flows that enable customers to build more secure and traceable AI applications.

1. Improved safety: Ability to filter out harmful content and unwanted topics in prompts and knowledge base nodes powered by Amazon Bedrock Guardrails. Guardrails are now supported on two types of nodes:
   • prompt node: Define and enforce controls over FM interactions.
   • knowledge base node: Apply guardrails to responses generated from your knowledge base.

2. Enhanced traceability: Ability to quickly validate and debug workflows with input/output traceability and inline validation. Gain comprehensive visibility into workflow execution and quickly identify errors through:
   • Support for detailed traceability of input and output nodes.
   • Complete execution path information showing inputs, outputs, execution time, and errors for each node.
   • Inline validation status of nodes in visual builder.

Consider ACME Corp, a fictitious e-commerce company that is building a customer service chatbot using Amazon Bedrock Flows. Implementation faces several challenges, including:

• The company’s chatbots sometimes generate responses that include sensitive customer information.
• They struggle to maintain consistent response quality and tone across different customer interactions.
• They spend a lot of time and effort troubleshooting application issues.
• There is no way to ensure that your answers comply with company policy or regulatory requirements.
• Lack of visibility into performance bottlenecks that impact customer experience.

Let’s take a look at how new features in Amazon Bedrock Flows address these challenges and enable Acme Corp to build more secure, efficient, and transparent customer service solutions.

Prerequisites

Before implementing new functionality, please ensure the following:

1. AWS account
2. Amazon Bedrock:
   • Create and test basic prompts for customer service interactions in Prompt Management.
   • Set up a knowledge base with relevant customer service documentation, FAQs, and product information.
   • Configure ancillary AWS services required for your customer service workflow, such as Amazon DynamoDB for order history.

3. With Amazon Bedrock Guardrails:
   • Create guardrail configurations for customer service interactions (e.g. CustomerServiceGuardrail-001) and:
     • Content filters for profanity and harmful content
     • Personally Identifiable Information (PII) Detection and Masking Rules for Customer Data
     • Custom word filter for company-specific terms
   • Contextual basis checks to ensure accurate information
   • Test and validate your guardrail configuration.
   • Publish the working version of the guardrail.

4. Required IAM permissions:

With these components in place, you can move on to implementing new functionality into your customer service workflow.

Enables enhanced flow security

For Acme Corp’s customer service chatbot, implementing guardrails ensures safe, compliant, and consistent customer interactions.

Here’s how to enable guardrails on both prompt nodes and knowledge base nodes:

1. In the AWS Management Console for Amazon Bedrock, open the Prompts node or the Knowledge Base node. customer service flow Where you want to add guardrails. Create new flows as needed.
   
2. In the node configuration panel, guardrail section.
   
   
3. Select an existing guardrail from the drop-down menu. for example, Customer Service Guardrail-001.
   
4. in this case, CustomerServiceGuardrail-001 is configured as follows:
   1. Mask customer PII data (name and email)
   2. Block profanity and harmful content
   3. Ask them to respond in accordance with company policy
   4. Maintain a professional tone in your responses

5. Choose the appropriate version of the guardrail. for example, working draft.
   
6. Enter the prompt message for your customer service scenario. for example, Respond to customer queries.
7. Connect the Prompt node to the flow’s input and output nodes.
   
8. Prompt the test flow to test the flow with the guardrails implemented. for example, Hi, my name is John Smith, email – john.smith@email.com. How do I get started with setting up an ACME Corp account?
   
9. in Test flow You can see how the model response handles sensitive information as shown in the right pane of the interface. for example:
   1. Original reply: “Dear Mr. John Smith…”
   2. Guardrail response: “Dear Mr. {name}…”

   

Enhance traceability with Flows Trace View

New flow tracing capabilities provide detailed visibility into flow execution and improved debugging capabilities with trace views and inline validation. This comprehensive monitoring solution helps developers more effectively monitor, debug, and optimize their AI workflows.

The main benefits of enhanced traceability are:

• Complete execution path with visibility through trace view
• Detailed input/output tracing for each node
• Errors, warnings, and execution timing for each node
• Quickly identify bottlenecks and issues
• Accelerate root cause analysis of errors

For Acme Corp’s customer service team, new flow tracing capabilities provide critical insight into chatbot performance and behavior. This helps with:

• Monitor response times for customer interactions
• Identify patterns in customer inquiries that cause delays
• Debugging conversation flow issues
• Optimize the customer experience

To use the trace view:

1. Open the flow in the Amazon Bedrock console and test it using a sample query.
   
2. After running the flow, select show trace Analyze interactions.
   
3. Check the flow trace window showing the following:
   1. Response time for each step of customer interaction
   2. How customer input is processed
   3. Where guardrails are installed
   4. performance bottleneck

   

4. Analyze execution details including:
   1. Customer query processing steps
   2. Generating and validating responses
   3. How long each step takes
   4. Error details and cause analysis

   

Inline validation status

The Flows visual builder and SDK now include intuitive node validation functionality.

Visual builder:

• A green background indicates a valid node configuration.
  
• A red background indicates an invalid node configuration that requires attention.
  
• A yellow background indicates a node configuration with warnings.
  

These validation features help developers quickly identify and resolve potential issues in their flows by providing real-time validation feedback during both visual and programmatic development.

conclusion

The integration of Bedrock Guardrails and enhanced traceability of Bedrock Flows represents a major advancement in generative AI development. These capabilities enable developers to create more secure, transparent, and efficient AI-powered solutions that address key challenges in the rapidly evolving field of AI application development.

Bedrock Flow with new features is now generally available in all regions where Amazon Bedrock is available, except GovCloud. Starting February 1, 2025, Bedrock Flows usage will also be charged at $0.035 per 1,000 node transitions, based on the number of node transitions required to operate the workflow. Explore these new features and experience first-hand how they can improve your generative AI development process. Get started by opening the Amazon Bedrock console and starting building secure, highly visible flows using Flows today. For more information, see the AWS User Guide for Guardrails Integration and Traceability. For pricing information, please visit the Amazon Bedrock pricing page.

We look forward to using these new features to build innovative applications. As always, we welcome your feedback through AWS re:Post for Amazon Bedrock or through your usual AWS contacts. Join the Community.aws Generative AI Builders community to share your experiences and learn from others.

About the author

Amit Lulla As a Principal Solutions Architect at AWS, I design enterprise-scale generative AI and machine learning solutions for software companies. With over 15 years of experience in software development and architecture, he is passionate about turning complex AI challenges into bespoke solutions that deliver real business value. When he’s not designing cutting-edge systems or mentoring fellow architects, you can find Amit practicing yoga on the squash court or planning his next travel adventure. He also practices meditation daily, which he believes keeps him focused in the fast-paced world of AI innovation.

Phuong Nguyen Principal Product Manager at AWS. She leads Amazon Bedrock Flows and has 18 years of experience building customer-centric, data-driven products. She is passionate about democratizing responsible machine learning and generative AI to enable customer experience and business innovation. Outside of work, I enjoy spending time with family and friends, listening to audiobooks, traveling, and gardening.