How MoEngage built agentic customer engagement on Amazon Bedrock AgentCore
MoEngage built Merlin AI, an agentic customer‑engagement platform on Amazon Bedrock AgentCore that lets marketers draft, analyze and adjust campaigns via a single conversational interface. The system uses Anthropic Claude Sonnet 4.6, 5 and 5.5 models for reasoning, tool‑call planning and content generation, and exposes existing MoEngage APIs as MCP‑compatible tools for secure, tenant‑scoped data access. A pilot with the Glance brand achieved a 38 % uplift in push‑notification click‑through rates. This demonstrates a reusable architecture for turning platform APIs into agent‑callable tools while preserving data isolation.
⚡ Key Takeaways
- 38 % uplift in push notification click‑through rates reported by Glance after using Merlin AI.
- Agents call existing MoEngage APIs exposed as MCP tools via the MCP Connector.
- Secure, tenant‑scoped data isolation is achieved without rebuilding hundreds of APIs.
- Uses Anthropic Claude Sonnet 4.6, 5, 5.5 on Amazon Bedrock for reasoning and generation.
- Hosted on Amazon Bedrock AgentCore runtime for scalable agent execution.
Engineers can rapidly add agentic capabilities to existing SaaS platforms by wrapping current APIs with MCP tools and leveraging Bedrock AgentCore, gaining measurable engagement gains without extensive API rewrites.
✅ Practical Steps
- Expose your platform’s existing APIs as MCP‑compatible tools using the MoEngage Model Context Protocol (MCP) Connector.
- Deploy agents on Amazon Bedrock AgentCore, selecting Anthropic Claude Sonnet models for reasoning and content generation.
Want the full story? Read the original article.
Read on AWS ML Blog ↗