Scaling agentic platforms using MCP and Amazon Bedrock AgentCore
In short: MoEngage has integrated AI agents into its customer engagement platform by using the Model Context Protocol (MCP) and Amazon Bedrock AgentCore to expose existing APIs as agent-callable tools. By treating its platform as an architecture of isolated sessions and standardized tool schemas, the company enabled teams to automate workflows and decision-making within a single conversation.
This summary was generated automatically by AI from AWS Machine Learning's publication. It is our own text, not a copy of the original — facts, figures and quotes belong to the source, linked above and below.
What changed?
- 1Converted existing OpenAPI 3.0-compliant platform APIs into MCP tools using the FastMCP framework.
- 2Implemented Amazon Bedrock AgentCore to handle session isolation, JWT authentication, and infrastructure scaling.
- 3Introduced Merlin AI agents capable of autonomous copy generation, flow management, and decisioning across communication channels.
Why it matters
This approach allows software providers to add agentic capabilities without rebuilding backend infrastructure. By decoupling the reasoning engine (models) from the data access layer (MCP server), you can maintain strict multi-tenant security and auditability while enabling complex cross-tool agent workflows.
What it means for AI agents and contact centers
For your work on voice agents and contact center automation, this pattern is a blueprint for exposing internal tools (like CRM lookups, call transcript analysis, or flow triggers) to LLMs securely. When testing, focus on how Bedrock AgentCore handles session isolation compared to custom proxy layers, and benchmark how the FastMCP approach compares to your existing tool-calling middleware in terms of latency and reliability.
🧪 Worth Testing
The architectural pattern of using FastMCP to automatically sync OpenAPI specs with agent tools significantly reduces maintenance overhead and is a robust way to integrate LLMs into production VoIP/SIP-related services.
Sources
- AWS Machine LearningOfficialPrimary sourceOriginal article →„How MoEngage built agentic customer engagement on Amazon Bedrock AgentCore“16 Nov 2026, 20:16
- Published by source
- 16 Nov 2026, 20:16
- Found by our system
- 16 Nov 2026, 20:36
- Summary generated
- 16 Nov 2026, 20:37
This article was written by AI from the original source. Facts, numbers and prices come from the source; missing values are marked “Not specified”. Legal notice, copyright and privacy