Coveo + AWS Bedrock Workshop¶
A hands-on builder's workshop exploring AI-powered search and conversational experiences
🎯 Objective¶
Master three integration patterns between Coveo and AWS Bedrock to build intelligent search and conversational AI solutions. This 90-minute hands-on workshop covers direct API integration, Bedrock Agent orchestration, and AgentCore with Model Context Protocol (MCP).
🏗️ What You Will Build¶
🔍 Lab 1: Direct Integration with Coveo API
Pattern: Coveo Direct API
Learn to integrate Coveo Search, Passage Retrieval, and Answer APIs directly into your application for intelligent search experiences.
⏱️ Duration: 20 minutes
🤖 Lab 2: Bedrock Agent with Coveo Tool
Pattern: Bedrock Agent
Configure AWS Bedrock Agent to use Coveo Passage Retrieval as a tool for grounded conversational AI responses.
⏱️ Duration: 20 minutes
⚡ Lab 3: AgentCore with Coveo MCP Server
Pattern: AgentCore Runtime + MCP
Deploy AWS Bedrock AgentCore with Model Context Protocol (MCP) for advanced agent orchestration with Coveo tools.
⏱️ Duration: 20 minutes
💬 Lab 4: Multi-Turn Conversations & Memory
Patterns: All Three Backends
Test conversational AI with session memory and cross-session recall using the chatbot interface across all integration patterns.
⏱️ Duration: 20 minutes
✅ Prerequisites¶
Required Access:
AWS Account Credentials
Your instructor will provide AWS account credentials and access instructions at the beginning of the workshop.
Search UI Account Credentials
Your instructor will provide Search UI credentials and access instructions at the beginning of the workshop.
- AWS Console access (Verify Access to AWS. NOTE: Switch to
us-east-1region) - Workshop UI Access (Verify Access to the search UI with the provided credentials)
Knowledge Base: Pre-indexed financial content from 11 authoritative sources.
View All Indexed Sources
- Wikipedia - General knowledge and financial concepts
- Investor.gov - Investment guidance and securities information
- IRS - Tax information and regulations
- NCUA - National Credit Union Administration resources
- FinCEN - Financial Crimes Enforcement Network guidance
- CFPB - Consumer Financial Protection Bureau resources
- FDIC - Federal Deposit Insurance Corporation information
- FRB - Federal Reserve Board policies and guidance
- OCC - Office of the Comptroller of the Currency regulations
- MyMoney.gov - Financial literacy and education resources
- FTC - Federal Trade Commission consumer protection guidance
All exercises are console-based - no command-line tools required.
Important: Model Throughput Limits
Due to the large number of workshop participants, the AWS Bedrock model throughput for this account may be temporarily exhausted. If you don't receive a response from the Bedrock model, please wait 30-60 seconds and retry your request. This is expected behavior during peak usage and does not indicate an error with your configuration.
🏗️ Deployed Infrastructure¶
Your AWS account includes pre-deployed components:
(App Runner)"] API["🔐 API Gateway
(HTTP API)"] AUTH["🔐 Cognito
(Authentication)"] L1["⚡ Search Proxy
(Lambda)"] L2["⚡ Passages Proxy
(Lambda)"] L3["⚡ Answer Proxy
(Lambda)"] L4["⚡ Agent Chat
(Lambda)"] L5["⚡ AgentCore Chat
(Lambda)"] BA["🤖 Bedrock Agent
(Action Groups)"] ACR["🤖 AgentCore Runtime
(Orchestrator)"] MCP["🤖 Coveo MCP Server
(Tool Runtime)"] MEM["🤖 Agent Memory
(Cross-Session)"] COVEO["🌐 Coveo Platform
(Search/Passages/Answer)"] UI -.->|Login| AUTH UI -->|HTTPS + JWT| API API -.->|Verify Token| AUTH API -->|/search| L1 API -->|/passages| L2 API -->|/answer| L3 API -->|/agent| L4 API -->|/agentcore| L5 L1 & L2 & L3 -->|Direct API| COVEO L4 -->|Invoke| BA L5 -->|Invoke| ACR BA -->|Tool Calls| COVEO ACR -->|MCP Protocol| MCP MCP -->|API Calls| COVEO BA -.->|Memory| MEM ACR -.->|Memory| MEM style UI fill:#e1f5fe,stroke:#01579b,stroke-width:3px,color:#000 style API fill:#f3e5f5,stroke:#4a148c,stroke-width:3px,color:#000 style AUTH fill:#f3e5f5,stroke:#4a148c,stroke-width:2px,color:#000 style L1 fill:#fff3e0,stroke:#e65100,stroke-width:2px,color:#000 style L2 fill:#fff3e0,stroke:#e65100,stroke-width:2px,color:#000 style L3 fill:#fff3e0,stroke:#e65100,stroke-width:2px,color:#000 style L4 fill:#fff3e0,stroke:#e65100,stroke-width:2px,color:#000 style L5 fill:#fff3e0,stroke:#e65100,stroke-width:2px,color:#000 style BA fill:#fce4ec,stroke:#880e4f,stroke-width:3px,color:#000 style ACR fill:#fce4ec,stroke:#880e4f,stroke-width:3px,color:#000 style MCP fill:#fce4ec,stroke:#880e4f,stroke-width:3px,color:#000 style MEM fill:#fce4ec,stroke:#880e4f,stroke-width:2px,color:#000 style COVEO fill:#e8f5e8,stroke:#1b5e20,stroke-width:3px,color:#000
Key Components:
- Search UI: Interactive interface for testing all integration patterns
- API Gateway + Cognito: Secure API access with JWT authentication
- Lambda Functions: Serverless proxies for each backend mode
- Bedrock Agent: AI orchestration with Coveo passage API tool integration
- AgentCore Runtime: Advanced agent platform with MCP protocol
- Coveo Platform: Enterprise search with AI-powered relevance
Workshop UI Features:
| Feature | Search Interface | Chatbot Interface |
|---|---|---|
| Core | Backend selection toggle • Search bar • Results with citations | Multi-turn conversations • Session memory |
| Display | AI-generated answers • Passage excerpts • Source filtering | Cross-session memory • Source attribution |
🚀 Progressive Learning Path¶
5 min] --> B[Lab 1
20 min] B --> C[Lab 2
20 min] C --> D[Lab 3
20 min] D --> E[Lab 4
20 min] E --> F[Q&A
5 min] style A fill:#e8f5e8 style B fill:#e1f5fe style C fill:#fff3e0 style D fill:#f3e5f5 style E fill:#fce4ec style F fill:#e8f5e8
| Lab | Duration | Focus |
|---|---|---|
| Lab 1 | 20 min | Direct Integration with Coveo API |
| Lab 2 | 20 min | Integrate Bedrock Agent with Coveo Passage Retrieval API Tool |
| Lab 3 | 20 min | Integrate Bedrock AgentCore with Coveo MCP Server |
| Lab 4 | 20 min | Test Multi-Turn Conversations with Agents |
Learning Objectives:
| Technical Skills | Business Understanding |
|---|---|
| 🔍 Master three Coveo-Bedrock integration patterns | ✅ Identify when to use each pattern |
| 🤖 Configure agents with custom tools and memory | ✅ Evaluate benefits and trade-offs |
| ⚡ Deploy AgentCore runtimes with MCP servers | ✅ Design case deflection strategies |
| 💬 Implement cross-session conversational memory | ✅ Assess ROI for intelligent search |
| 📊 Observe agent behavior through AWS tooling | ✅ Apply production-ready patterns |
🎉 Let's Get Started!¶
Workshop Support
If you encounter issues, ask your instructor for assistance.