Build WhatsApp AI agents with Amazon Bedrock AgentCore
SMART, an open-source Python SDK and Terraform stack, deploys the full WhatsApp channel layer to your AWS account with one command. Bring any agent that runs on Amazon Bedrock AgentCore, and SMART handles webhook ingestion, media, conversation memory, scheduling, and Meta message templates.
WhatsApp is one of the most widely used messaging apps across Latin America, India and Southeast Asia. This makes it a natural choice for building AI agents. These agents, running on Amazon Bedrock AgentCore, can provide a natural interaction surface for users. However, building integration for such agents is challenging. Inbound messages are deeply nested payloads that need to be authenticated and normalized.
Media such as voice notes, images, and documents each require their own processing pipeline. Conversation memory that survives sessions, durable schedules for reminders, and an authenticated path for outbound messages also need to be handled. Anything sent more than 24 hours after the last user message requires a pre-approved Meta message template.
Integrating all these components requires configuring over 10 AWS services and understanding WhatsApp system behaviors, which most teams discover only by using them. To simplify this integration, SMART (Social Messaging Agent Runtime Toolkit) offers an open source Python SDK and Terraform stack. With just one command, it deploys the entire infrastructure to the user's AWS account.
Teams are already using SMART with Amazon Bedrock AgentCore. Whether starting from scratch or migrating an existing agent, teams own the agent's logic while SMART handles the channel layer. The table below compares building the channel from scratch versus using SMART. Concern Building it yourself With SMART Webhook ingestion Ingestion, signature verification, nested payloads, a dozen message types Handled, clean prompt Media Download, transcribe, store, cleanup, outbound file upload Full lifecycle, both directions Message templates and 24-hour window Meta approval flow, category rules, conversation stage Declared in YAML, synced by CLI Conversation memory Multi-tier memory, session handling Amazon Bedrock AgentCore memory, one function call Scheduling and triggers EventBridge wiring, authenticated ingress, routing Infrastructure More than 10 AWS services One terraform apply Local development Redeploy and tunnel on every change Test locally without redeploying The agentic layer Amazon Bedrock AgentCore platform with AgentCore runtime hosts and scales agents without managing infrastructure.
AgentCore memory persists conversation history across sessions and deployments. Because AgentCore is framework-agnostic, SMART builds on this capability. Strands Agents integrates conversation memory automatically across frameworks. LangGraph, Pydantic-AI, or other frameworks can integrate with SMART. The SDK exposes memory and session hooks regardless of framework.
Teams keep control over the elements that differentiate their product: framework, model, prompt, and business logic.
Written by urgent.news from AWS Machine Learning's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.