Connecting Pega to MCP Servers and Conversational AI Agents
Purpose
The TeeNow application is built to demonstrate how a Pega Constellation application can be integrated with MCP (Model Context Protocol) servers and conversational AI agents to deliver real-time, natural-language experiences to end users.
Using a golf tee-time booking scenario, TeeNow shows how a Pega application can host a conversational AI agent that talks to an external MCP server in plain English — letting users ask questions like “Find golf courses in Wisconsin with a spa” and get live answers, without writing a single line of SQL or navigating a search form.
What TeeNow Showcases
- A modern Pega Constellation portal with an AI assistant on the home screen as the primary way users interact with the application.
- A conversational AI agent (TeeNowAgent), powered by Claude Sonnet 4.6 via AWS Bedrock, that understands natural-language questions about golf courses.
- A live MCP server integration that lets the AI agent fetch and search golf course data from an external service in real time, rather than from a static list inside Pega.
- A foundation for booking workflows — case types for tee-time booking, course management, reservations, payments, notifications, and platform onboarding are all set up and ready to be wired into the conversational experience.
How the Pieces Connect
User
↓
TeeNow Portal (Pega Constellation)
↓
TeeNow AI Assistant (Claude Sonnet 4.6)
↓
MCP Connector (Streamable HTTP)
↓
External Golf Courses MCP Server
↓
Live Course Data (with search & filtering)
The key idea: Pega owns the user experience and the business workflows; the MCP server owns the data and the search logic. The two are loosely coupled through the standard MCP protocol, so the same Pega application could swap in a different MCP server without changing its UI or its case types.
Related Documentation
This overview is intentionally light. For deeper detail, see the companion documents: (check the 1st reply for the following documents) 
-
User Guide — How to use the application: opening the portal, asking the AI assistant for courses, and what to expect from the conversational experience. -
TeeNow Golf Booking Application integrated to a MCP server — The technical implementation document: the Pega rules involved (Portal, AI Agent, MCP Connector), configuration details, model setup, and how the components reference each other. -
Build MCP Server Readme — How the external MCP server itself was built and deployed on a serverless edge platform — including the database schema, the prompt that scaffolded it, and the steps used to make it available to AI assistants.
Status
TeeNow is an active demonstration project. The core integration — Pega portal ↔ AI agent ↔ MCP server — is working end to end and the AI assistant successfully lists and searches golf courses from the live external service. Some refinements (system prompt, guardrails, response styling) are still being configured.
access TeeNow@ Url : https://lab-18233-us-east-1.internal.pegalabs.io/prweb/app/tee-now
user/pass : xxxx
Next on the roadmap: extending the conversational experience to invoke a Pega case to complete an actual tee-time booking once a course is chosen.