Scaling shopping help on WhatsApp was too expensive. Now, layered AI handles 100k+ hourly messages in local dialects.
A technology startup serving the Latin American market, where messaging platforms like WhatsApp drive roughly 70% of all commercial transactions.
The company aimed to build a multimodal assistant capable of understanding local Spanish and Portuguese dialects while handling complex purchasing...
“Claude on Vertex AI proved to be the right choice very early on – in a simple A/B test with users right after deploying Claude 3 Haiku, we saw an important increase in ‘thumbs up’ reactions to Zapia interactions, to well over 90%.”
WhatsApp-based AI agent for real-world tasks in Latin American markets.
Cloud computing services, AI infrastructure, and data analytics platforms for enterprises.
Related implementations across industries and use cases
Manual chats slowed commerce. Now, a dialect-aware AI handles 100k+ WhatsApp messages an hour for 2.5M users.
Language barriers hid 2M products from global buyers. AI now rewrites queries to find exact matches across borders.
Legacy tools failed on nuanced emails. Now, a network of agents reasons through complex requests, automating 71% of support.
Manual fixes slowed order processing to 2.5 hours. AI code optimization cut that to 30 minutes and reduced incidents by 50%.
300+ receipt layouts forced manual entry. GenAI now reads raw images with 98.6% accuracy, cutting costs 90%.
Lab supply orders were handwritten in notebooks. Digital ordering now takes seconds, saving 30,000 hours for research annually.
Experts spent 15 minutes pulling data from scattered systems. Natural language prompts now generate detailed reports instantly.