Reactiv Enhances Mobile Commerce Efficiency with Amazon Bedrock AgentCore
Reactiv has introduced a mobile commerce solution that empowers Shopify merchants to efficiently launch and manage native mobile applications. These apps are crucial as they reportedly convert shoppers at rates 2 to 4 times higher than traditional web visitors. However, keeping these applications updated can be labor-intensive, requiring merchants to manually select featured products, rearrange content, and publish updates regularly. Recognizing the challenges faced by merchants, Reactiv leveraged Amazon Bedrock AgentCore to automate these processes, leading to an impressive 80% reduction in configuration time and a 33% faster path to production.
The company developed an AI Scheduler that autonomously refreshes merchant apps based on predefined schedules. Merchants can simply express their needs in natural language, such as "Refresh my homepage with best sellers every Monday at 9 AM," and the system manages the rest. This innovation was realized through a three-agent architecture built on Amazon Bedrock AgentCore, utilizing the Strands Agents SDK, and was operational within weeks. By unifying their interactive and scheduled agents onto a single framework, Reactiv achieved shared memory across both functionalities, enhancing efficiency.
Reactiv specializes in creating native iOS and Android applications for Shopify merchants, equipped with a low-code app builder and analytics tools. Prior to the AI Scheduler, the company had developed a conversational AI tool that allowed merchants to interact with an AI assistant in real-time to modify their applications. However, merchants expressed a desire for an autonomous solution that could handle updates without any manual input. This necessitated a more sophisticated architecture than what was previously available.
The solution required capabilities such as hosting multi-agent systems, maintaining persistent memory across sessions, and ensuring automatic data isolation for each merchant. Amazon Bedrock AgentCore provides a robust platform for building and optimizing agents at scale, addressing these requirements effectively. Key features of AgentCore include managed agent execution, persistent memory, and service-to-service authentication, all of which Reactiv utilized to enhance their system.
AgentCore's runtime operates agents in Firecracker microVMs, which offer isolation similar to AWS Lambda. Reactiv packages their agent code as Docker images, allowing for seamless deployment and execution. This architecture eliminates the need for managing containers or scaling policies, simplifying the deployment process significantly. Furthermore, AgentCore's built-in memory capabilities ensure that each merchant's data remains private and that the system can learn from past interactions to improve future performance.
Streamlining Merchant App Configuration
The integration of AgentCore also allows Reactiv to host their Config MCP natively, which initializes the merchant's app configuration at the start of each session. This setup enables the Builder Agent to perform real-time updates while ensuring compliance with the established schema. By leveraging AgentCore Identity, Reactiv streamlined the authentication process, removing the need for custom layers that were previously necessary.
The architecture also supports multi-tenant isolation, ensuring that each merchant's data and execution context remain distinct and secure. This is achieved through Firecracker microVMs, which facilitate independent execution environments for each merchant's agents. As a result, merchants can confidently utilize the system without concerns about data leakage or cross-merchant interference.
Why It Matters
With the launch of the AI Scheduler, Reactiv transitioned from a single-agent system to a unified multi-agent framework on Amazon Bedrock AgentCore. This evolution has resulted in a more efficient production system that incorporates three specialized agents and over 50 tools, all sharing persistent memory across interactive and autonomous modes. The reusable architecture has proven beneficial, allowing Reactiv to extend its capabilities to new tools and services.
The practical implications of this advancement are significant. By enabling merchants to customize their mobile apps through natural language, Reactiv is enhancing the user experience and driving engagement. Future developments may include allowing merchants to request entirely new layout sections via natural language, further expanding the agent's capabilities while adhering to brand guidelines. Reactiv's journey from self-managed infrastructure to a sophisticated, unified system illustrates the potential of AI in transforming mobile commerce for Shopify merchants.




