This post, co-authored with Cyril Ovely from Vxceed, addresses a pressing challenge for consumer packaged goods (CPG) companies in emerging economies: effectively retaining revenue and growing customer loyalty at scale. Despite investing 15–20% of their revenue in trade promotions and retailer loyalty programs, uptake of these initiatives remains below 30% due to their complexity and difficulties in meeting individual retailer needs. Vxceed’s Lighthouse platform meets this challenge with its innovative loyalty module, trusted by leading global CPG brands across Southeast Asia, Africa, and the Middle East. Equipped with an AI-driven toolkit powered by generative AI, Lighthouse enables field sales teams to craft personalized sales pitches tailored to individual retailers’ data and trends. This personalization enhances retailer engagement, overcomes objections, and increases program adoption. This article details how Vxceed leveraged Amazon Bedrock—a fully managed service offering access to multiple high-performing foundation models (FMs) via a single API—to build their AI-powered multi-agent solution that generates personalized sales pitches at scale. **The Challenge: Enhancing Revenue Retention in Emerging Markets** Operating primarily in emerging economies, Vxceed contends with an industry marked by rapid change, high customer expectations, and low entry barriers. CPG companies worldwide allocate 15–20% of revenues annually to trade promotions and loyalty programs, mostly sold through field sales teams who manage millions of retail outlets. However, uptake rates often fall below 30% due to program complexity and the necessity to tailor loyalty offerings to each outlet. To improve adoption and thereby boost revenue retention, loyalty programs require personalized pitches that suit each retailer’s unique needs. Vxceed needed a scalable solution generating customized selling stories for field sales teams to effectively promote these programs. **Solution Overview: Lighthouse Loyalty Selling Story** Vxceed developed Lighthouse Loyalty Selling Story, an AI-powered solution leveraging Amazon Bedrock, Amazon API Gateway, Amazon DynamoDB, and AWS Lambda to deliver a secure, scalable, and serverless system. The architecture employs a multi-agent framework within the customer’s AWS environment, ensuring data security and seamless user experience. Key components include: - **Salesperson App:** A mobile interface through which field sales teams access tailored sales pitches and interact via chat. - **API Gateway and Security:** Manages secure access with AWS Key Management Service and Secrets Manager, while Amazon S3 handles image storage. - **Intelligent Agents (Lambda functions):** - *Orchestration Agent* coordinates workflows. - *Story Framework Agent* defines narrative structures. - *Story Generator Agent* creates personalized content using outlet profiles, loyalty program, and historical data. - *Story Review Agent* validates content quality and compliance. - *Brand Guidelines Agent* ensures adherence to brand voice and standards. - *Business Rules Agent* applies operational and compliance rules. - **Data Services Layer:** Provides critical data via APIs—including outlet profiles, loyalty details, purchase history—and integrates with AI/ML models and data lakes. - **Advanced Capabilities:** Includes Q&A services for natural language queries, streamlined signup processes via call-to-action services, AI-powered responses from Amazon Bedrock LLMs, and compliance-focused guardrails. **Multi-Agent AI Architecture** The multi-agent system orchestrates personalized storytelling through specialized Lambda functions that collaborate with the Amazon Bedrock LLM.

Each agent contributes distinct functions, from content generation to brand consistency and enforcing business logic, ensuring scalable, cost-effective, and secure AI-driven interactions. **Guardrails** Amazon Bedrock Guardrails enforce appropriate, professional communication by filtering denied topics and offensive language, preventing off-topic or unprofessional discourse and managing sensitive or competitive inquiries tactfully in line with corporate values. **Why Amazon Bedrock?** Vxceed’s choice of Amazon Bedrock was based on: 1. **Enterprise-grade security and privacy:** AI workloads and data remain confined within customers’ virtual private clouds, under encryption and isolated. 2. **Managed AWS services integration:** Enables seamless operation within Vxceed’s existing AWS infrastructure. 3. **Access to multiple AI foundation models:** Allowing optimization across use cases, with Anthropic’s Claude 3. 5 Sonnet favored for sophisticated conversational capabilities. 4. **Robust AI development tools:** Including knowledge bases, prompt engineering libraries, and agent frameworks for efficient AI orchestration. **Business Impact and Future Outlook** The solution realized marked improvements: - Achieved 95% response accuracy and automated 90% of loyalty program queries, enhancing customer service and retailer experience. - Increased program enrollments by 5–15%, demonstrating direct revenue growth impact. - Delivered operational efficiencies such as 20% faster enrollment processing, 10% reduced support time, and saved two person-months annually per region. Looking ahead, Vxceed plans to optimize AI inference costs for scalability, incorporate language agents for native language presentations, and employ Retrieval-Augmented Generation (RAG) and GraphRAG techniques to refine story generation. **Conclusion** Through collaboration with AWS, Vxceed successfully developed Lighthouse Loyalty Selling Story, an AI-powered, scalable, and secure solution that personalizes sales pitches for CPG field teams in emerging markets. This innovation elevates revenue retention, operational efficiency, and customer engagement. Organizations interested in similar AI deployments can explore asynchronous AI agent implementation using Amazon Bedrock to build secure, multi-agent AI solutions. **About the Authors** [Author details would be provided here. ]