AI agents are revolutionizing how businesses enhance their operational capabilities and enterprise applications. By enabling natural language interactions, these agents provide customers with a streamlined, personalized experience. Amazon Bedrock Agents uses the capabilities of foundation models (FMs), combining them with APIs and data to process user requests, gather information, and execute specific tasks effectively. The introduction of multi-agent collaboration now enables organizations to orchestrate multiple specialized AI agents working together to tackle complex, multi-step challenges that require diverse expertise.
Amazon Bedrock offers a diverse selection of FMs, allowing you to choose the one that best fits your specific use case. Among these offerings, Amazon Nova stands out as AWS’s next-generation FM, delivering breakthrough intelligence and industry-leading performance at exceptional value.
The Amazon Nova family comprises three types of models:
Understanding models – Available in Micro, Lite, and Pro variants
Content generation models – Featuring Canvas and Reel
Speech-to-Speech model – Nova Sonic
These models are specifically optimized for enterprise and business applications, excelling in the following capabilities:
Text generation
Summarization
Complex reasoning tasks
Content creation
This makes Amazon Nova ideal for sophisticated use cases like our FinOps solution.
A key advantage of the Amazon Nova model family is its industry-leading price-performance ratio. Compared to other enterprise-grade AI models, Amazon Nova offers comparable or superior capabilities at a more competitive price point. This cost-effectiveness, combined with its versatility and performance, makes Amazon Nova an attractive choice for businesses looking to implement advanced AI solutions.
In this post, we use the multi-agent feature of Amazon Bedrock to demonstrate a powerful and innovative approach to AWS cost management. By using the advanced capabilities of Amazon Nova FMs, we’ve developed a solution that showcases how AI-driven agents can revolutionize the way organizations analyze, optimize, and manage their AWS costs.
Solution overview
Our innovative AWS cost management solution uses the power of AI and multi-agent collaboration to provide comprehensive cost analysis and optimization recommendations. The core of the system is built around three key components:
FinOps supervisor agent – Acts as the central coordinator, managing user queries and orchestrating the activities of specialized subordinate agents
Cost analysis agent – Uses AWS Cost Explorer to gather and analyze cost data for specified time ranges
Cost optimization agent – Uses the AWS Trusted Advisor Cost Optimization Pillar to provide actionable cost-saving recommendations
The solution integrates the multi-agent collaboration capabilities of Amazon Bedrock with Amazon Nova to create an intelligent, interactive, cost management AI assistant. This integration enables seamless communication between specialized agents, each focusing on different aspects of AWS cost management. Key features of the solution include:
User authentication through Amazon Cognito with role-based access control
Frontend application hosted on AWS Amplify
Real-time cost insights and historical analysis
Actionable cost optimization recommendations
Parallel processing of tasks for improved efficiency
By combining AI-driven analysis with AWS cost management tools, this solution offers finance teams and cloud administrators a powerful, user-friendly interface to gain deep insights into AWS spending patterns and identify cost-saving opportunities.
The architecture displayed in the following diagram uses several AWS services, including AWS Lambda functions, to create a scalable, secure, and efficient system. This approach demonstrates the potential of AI-driven multi-agent systems to assist with cloud financial management and solve a wide range of cloud management challenges.
In the following sections, we dive deeper into the architecture of our solution, explore the capabilities of each agent, and discuss the potential impact of this approach on AWS cost management strategies.
Prerequisites
You must have the following in place to complete the solution in this post:
Deploy solution resources using AWS CloudFormation
This CloudFormation template is designed to run in the us-east-1 Region. If you deploy in a different Region, you must configure cross-Region inference profiles to have proper functionality and update the CloudFormation template accordingly.
During the CloudFormation template deployment, you will need to specify three required parameters:
Stack name
FM selection
Valid user email address
AWS resource usage will incur costs. When deployment is complete, the following resources will be deployed:
Amazon Cognito resources:
AWS Identity and Access Management (IAM) resources:
IAM roles:
FinanceUserRestrictedRole
DefaultCognitoAuthenticatedRole
IAM policies:
Finance-BedrockAccess
Default-CognitoAccess
Lambda functions:
TrustedAdvisorListRecommendationResources
TrustedAdvisorListRecommendations
CostAnalysis
ClockandCalendar
CostForecast
Amazon Bedrock agents:
FinOpsSupervisorAgent
CostAnalysisAgent with action groups:
CostAnalysisActionGroup
ClockandCalendarActionGroup
CostForecastActionGroup
CostOptimizationAgent with action groups:
TrustedAdvisorListRecommendationResources
TrustedAdvisorListRecommendations
After you deploy the CloudFormation template, copy the following from the Outputs tab on the AWS CloudFormation console to use during the configuration of your application after it’s deployed in Amplify:
AWSRegion
BedrockAgentAliasId
BedrockAgentId
BedrockAgentName
IdentityPoolId
UserPoolClientId
UserPoolId
The following screenshot shows you what the Outputs tab will look like.
Deploy the Amplify application
You need to manually deploy the Amplify application using the frontend code found on GitHub. Complete the following steps:
Download the frontend code AWS-Amplify-Frontend.zip from GitHub.
Use the .zip file to manually deploy the application in Amplify.
Return to the Amplify page and use the domain it automatically generated to access the application.
Amazon Cognito for user authentication
The FinOps application uses Amazon Cognito user pools and identity pools to implement secure, role-based access control for finance team members. User pools handle authentication and group management, and identity pools provide temporary AWS credentials mapped to specific IAM roles. The system makes sure that only verified finance team members can access the application and interact with the Amazon Bedrock API, combining robust security with a seamless user experience.
Amazon Bedrock Agents with multi-agent capability
The Amazon Bedrock multi-agent architecture enables sophisticated FinOps problem-solving through a coordinated system of AI agents, led by a FinOpsSupervisorAgent. The FinOpsSupervisorAgent coordinates with two key subordinate agents: the CostAnalysisAgent, which handles detailed cost analysis queries, and the CostOptimizationAgent, which handles specific cost optimization recommendations. Each agent focuses on their specialized financial tasks while maintaining contextual awareness, with the FinOpsSupervisorAgent managing communication and synthesizing comprehensive responses from both agents. This coordinated approach enables parallel processing of financial queries and delivers more effective answers than a single agent could provide, while maintaining consistency and accuracy throughout the FinOps interaction.
Lambda functions for Amazon Bedrock action groups
As part of this solution, Lambda functions are deployed to support the action groups defined for each subordinate agent.
The CostAnalysisAgent uses three distinct Lambda backed action groups to deliver comprehensive cost management capabilities. The CostAnalysisActionGroup connects with Cost Explorer to extract and analyze detailed historical cost data, providing granular insights into cloud spending patterns and resource utilization. The ClockandCalendarActionGroup maintains temporal precision by providing current date and time functionality, essential for accurate period-based cost analysis and reporting. The CostForecastActionGroup uses the Cost Explorer forecasting function, which analyzes historical cost data and provides future cost projections. This information helps the agent support proactive budget planning and make informed recommendations. These action groups work together seamlessly, enabling the agent to provide historical cost analysis and future spend predictions while maintaining precise temporal context.
The CostOptimizationAgent incorporates two Trusted Advisor focused action groups to enhance its recommendation capabilities. The TrustedAdvisorListRecommendationResources action group interfaces with Trusted Advisor to retrieve a comprehensive list of resources that could benefit from optimization, providing a targeted scope for cost-saving efforts. Complementing this, the TrustedAdvisorListRecommendations action group fetches specific recommendations from Trusted Advisor, offering actionable insights on potential cost reductions, performance improvements, and best practices across various AWS services. Together, these action groups empower the agent to deliver data-driven, tailored optimization strategies by using the expertise embedded in Trusted Advisor.
Amplify for frontend
Amplify provides a streamlined solution for deploying and hosting web applications with built-in security and scalability features. The service reduces the complexity of managing infrastructure, allowing developers to concentrate on application development. In our solution, we use the manual deployment capabilities of Amplify to host our frontend application code.
Multi-agent and application walkthrough
To validate the solution before using the Amplify deployed frontend, we can conduct testing directly on the AWS Management Console. By navigating to the FinOpsSupervisorAgent, we can pose a question like “What is my cost for Feb 2025 and what are my current cost savings opportunity?” This query demonstrates the multi-agent orchestration in action. As shown in the following screenshot, the FinOpsSupervisorAgent coordinates with both the CostAnalysisAgent (to retrieve February 2025 cost data) and the CostOptimizationAgent (to identify current cost savings opportunities). This illustrates how the FinOpsSupervisorAgent effectively delegates tasks to specialized agents and synthesizes their responses into a comprehensive answer, showcasing the solution’s integrated approach to FinOps queries.
Navigate to the URL provided after you created the application in Amplify. Upon accessing the application URL, you will be prompted to provide information related to Amazon Cognito and Amazon Bedrock Agents. This information is required to securely authenticate users and allow the frontend to interact with the Amazon Bedrock agent. It enables the application to manage user sessions and make authorized API calls to AWS services on behalf of the user.
You can enter information with the values you collected from the CloudFormation stack outputs. You will be required to enter the following fields, as shown in the following screenshot:
User Pool ID
User Pool Client ID
Identity Pool ID
Region
Agent Name
Agent ID
Agent Alias ID
Region
You need to sign in with your user name and password. A temporary password was automatically generated during deployment and sent to the email address you provided when launching the CloudFormation template. At first sign-in attempt, you will be asked to reset your password, as shown in the following video.
Now you can start asking the same question in the application, for example, “What is my cost for February 2025 and what are my current cost savings opportunity?” In a few seconds, the application will provide you detailed results showing services spend for the particular month and savings opportunity. The following video shows this chat.
You can further dive into the details you got by asking a follow-up question such as “Can you give me the details of the EC2 instances that are underutilized?” and it will return the details for each of the Amazon Elastic Compute Cloud (Amazon EC2) instances that it found underutilized.
The following are a few additional sample queries to demonstrate the capabilities of this tool:
What is my top services cost in June 2024?
In the past 6 months, how much did I spend on VPC cost?
What is my current savings opportunity?
Clean up
If you decide to discontinue using the FinOps application, you can follow these steps to remove it, its associated resources deployed using AWS CloudFormation, and the Amplify deployment:
Delete the CloudFormation stack:
On the AWS CloudFormation console, choose Stacks in the navigation pane.
Locate the stack you created during the deployment process (you assigned a name to it).
Select the stack and choose Delete.
Delete the Amplify application and its resources. For instructions, refer to Clean Up Resources.
Considerations
For optimal visibility across your organization, deploy this solution in your AWS payer account to access cost details for your linked accounts through Cost Explorer.
Trusted Advisor cost optimization visibility is limited to the account where you deploy this solution. To expand its scope, enable Trusted Advisor at the AWS organization level and modify this solution accordingly.
Before deploying to production, enhance security by implementing additional safeguards. You can do this by associating guardrails with your agent in Amazon Bedrock.
Conclusion
The integration of the multi-agent capability of Amazon Bedrock with Amazon Nova demonstrates the transformative potential of AI in AWS cost management. Our FinOps agent solution showcases how specialized AI agents can work together to deliver comprehensive cost analysis, forecasting, and optimization recommendations in a secure and user-friendly environment. This implementation not only addresses immediate cost management challenges, but also adapts to evolving cloud financial operations. As AI technologies advance, this approach sets a foundation for more intelligent and proactive cloud management strategies across various business operations.
Additional resources
To learn more about Amazon Bedrock, refer to the following resources:
About the Author
Salman Ahmed is a Senior Technical Account Manager in AWS Enterprise Support. He specializes in guiding customers through the design, implementation, and support of AWS solutions. Combining his networking expertise with a drive to explore new technologies, he helps organizations successfully navigate their cloud journey. Outside of work, he enjoys photography, traveling, and watching his favorite sports teams.
Ravi Kumar is a Senior Technical Account Manager in AWS Enterprise Support who helps customers in the travel and hospitality industry to streamline their cloud operations on AWS. He is a results-driven IT professional with over 20 years of experience. In his free time, Ravi enjoys creative activities like painting. He also likes playing cricket and traveling to new places.
Sergio Barraza is a Senior Technical Account Manager at AWS, helping customers on designing and optimizing cloud solutions. With more than 25 years in software development, he guides customers through AWS services adoption. Outside work, Sergio is a multi-instrument musician playing guitar, piano, and drums, and he also practices Wing Chun Kung Fu.
Ankush Goyal is a Enterprise Support Lead in AWS Enterprise Support who helps customers streamline their cloud operations on AWS. He is a results-driven IT professional with over 20 years of experience.
