Learn how a leading QSR chain leveraged Amazon Bedrock and serverless AI to boost customer conversion by 79%.
¡Hola! Me pasás una grande de muzza, mitad sin cebolla, y una gaseosa de 1.5? Es para ver el partido
Dale, confirmo
¡Listo! Tu pedido ya está en preparación y sumaste puntos para el próximo partido.
Illustrative conversation: order capture, regional food modifiers and geolocation routing in a single chat.
conversion rate from sessions to registered leads.
precision rate in natural language intent parsing.
reduction in customer acquisition and qualification costs.
Hornero is a leading quick-service restaurant (QSR) and pizzeria chain in Latin America. Operating dozens of physical locations since 1978, the company serves millions of customers annually, managing high seasonal traffic spikes, complex retail transactions, and dynamic tax compliance.
biia’s Generative AI solution transformed our digital commerce, enabling seamless ordering during peak match days while slashing customer acquisition costs by 75%.
Hornero collaborated with AWS Partner biia Group LLC to eliminate digital friction and replace a manual customer support operation with a high-availability conversational e-commerce and fan engagement platform. The client faced high shopping cart abandonment and lead leakage during peak operational hours. biia Group engineered a serverless, machine learning solution to automate order capture, process complex regional food modifiers, and execute instant cart recovery workflows. The production platform delivered a 79% lead conversion rate, 94% language parsing accuracy, and reduced customer acquisition costs by 75% while maintaining absolute system stability.
The client faced severe operational bottlenecks due to a 100% manual customer service operation across digital channels, leading to high cart abandonment rates, heavy workforce saturation during peak hours, and massive lead leakage outside business hours. Disconnected data silos between their physical point-of-sale systems, mobile apps, and customer platforms hindered targeted marketing automation and restricted real-time visibility into customer buying behavior.
To support high-revenue sales channels, the company required an elastic architecture capable of handling hundreds of thousands of concurrent transactional conversations, parsing natural language for variable food modifiers and dynamic tax structures, and routing orders based on customer geolocation. They needed an enterprise-grade software solution to guarantee continuous runtime availability, proactive security operations, and zero data loss during extreme traffic surges.
AWS Partner biia Group LLC architected a serverless Generative AI conversational platform deployed using Infrastructure as Code. The solution utilizes Amazon Bedrock to orchestrate specialized foundation models fine-tuned to process colloquial regionalisms, product synonyms, and intent classification. To ensure operational safety, Amazon Bedrock Guardrails was integrated to prevent non-compliant responses, enforce contextual grounding, and safeguard end-user transactional data.
The compute layer relies on AWS Lambda functions to process dynamic JSON payloads, update menus in real-time, compute dynamic taxation, and trigger payment gateways. Amazon DynamoDB acts as the centralized stateful engine, capturing complete conversation histories, keeping carts active across volatile connections, and triggering cart-recovery workflows. Document assets and logs are stored securely in Amazon S3 with automated lifecycle policies and encryption.
AI that understands the order, its changes and how the customer speaks.
Order payment straight from the conversation.
Order submission and order status tracking.
The production deployment achieved exceptional performance, delivering a 79.24% conversion rate from unique conversational sessions to registered leads (6,719 leads generated out of 8,479 conversations across 18,867 sessions). The platform maintained a 94% precision rate in natural language intent parsing (5,110 purchase receipts processed and 25,681 match predictions generated) while ensuring a 99.20% message delivery effectiveness rate and 99.9% architectural stability during extreme traffic spikes, where 40.76% of weekly traffic occurred during national team match days.
The system processed 126,457 engagement points ($31,614 USD value), issuing 19,233 discount coupons and driving over 1,070 direct point redemptions ($16,206 USD redeemed in-store) with zero state loss. biia Group’s serverless design reduced customer acquisition and lead qualification costs by 75% compared to manual campaigns. By unifying user context across physical branches (40.76% QR capture) and delivery channels (31.45%), the solution achieved an 80.10% active participation rate while driving $64,200 USD in annual recurring cloud revenue.
Every order, coupon and lead in this story started with a chat. Tell us about your channels and we’ll show you what biiaScale™ can do with them.
Talk to biia
