Agentic AI Implementation

AWS will advise and assist Customer in designing and developing an agentic artificial intelligence (AgenticAI) solution designed to [insert AWS-approved Customer use case] as set forth in more detail below (the “Solution”).

AWS will advise and assist Customer, in a non-production environment with the following:

• Design integration with Application Programming Interface’s (APIs)
• Design business process flow and associated data formats
• Define data integration APIs with underlying data lake and other necessary data sources
• Design end to end agentic architecture considering tools and/or MCP servers and API's
• Conduct prompt engineering
• Document Key Performance Indicators (KPIs) and expected ground truth data and results
• Design Agent platform including Tools and / or MCP server and relevant documentation
• Define Security and Networking architecture

AWS will provide the following deliverables to Customer during the engagement (“Deliverables”):

• Well-architected Agent Factory Framework which enables deployment, monitoring, and configuration of agents.

AWS will advise and assist Customer, in a non-production environment with the following:

• Design Agent Vending framework
• Compose API/CLI based agent framework code
• Define use of Agent framework such as Strands
• Implement RESTful APIs for agent blueprint creation, publishing, and management
• Define use of Command Line Interface (CLI) tool for developers to create and publish blueprints
• Deliver API-based marketplace with search and filtering capabilities
• Support deployment to Amazon Bedrock AgentCore

AWS will provide the following deliverables to Customer during the engagement (“Deliverables”):

• Agent Vending Solution

• Implement Data retrieval through underlying data storage using data APIs / tools
• Implement business logic within Agent to fetch the data required for response or analysis.
• Define and integrate Bedrock Guardrails
• Conduct and reiterate prompt engineering fundamentals for accurate agent responses and / or actions finalizing the models
• Implement security and networking guardrails

AWS will provide the following deliverables to Customer during the engagement (“Deliverables”):

• Deploy Agents hosted on AgentCore with well architected APIs, tools and /or MCP servers on AWS

AWS will advise and assist Customer, in a non-production environment with the following:

• Validate agent factory framework and agent performance against goals identified during discovery
• Conduct knowledge transfer to customer team
• Review Documentation serving as a knowledge base for deployed agents and underlying platform
• Review Documentation for scaling pipelines to handle customer-defined production workloads
• Provide training and documentation enabling customer teams to independently operate and scale the platform

AWS will provide the following deliverables to Customer during the engagement (“Deliverables”):

• Demo platform with validated deployment and integration
• Operational documentation