Vambe powers conversational commerce across Latin America with Claude

Vambe is a Chilean startup building conversational AI that helps mid-sized businesses across Latin America. The platform turns customer interactions into revenue by automating sales conversations, qualifying leads, and completing transactions directly within WhatsApp and Instagram.

With Claude, Vambe achieved:

• 95%+ multi-agent reliability, up from 30%, enabling multi-agent workflows
• 40-60% higher conversion rates compared to customers' previous sales processes, including both manual workflows and prior AI models
• 60% reduction in healthcare clinic no-shows
• Customer onboarding time reduced from 7 days to 3 days
• Engineering debugging time dropped from 40% to 10%, freeing 120 hours per month 

The challenge

Mid-sized businesses across Latin America face a common challenge: top-tier enterprise automation tools are too expensive and complex, while basic software cannot keep pace with growth. Sales teams spend entire days answering repetitive questions—"What are your hours?" "Do you have this in stock?" "How much does it cost?"—instead of closing deals.

Vambe set out to solve this with a platform that qualifies leads, schedules appointments, and completes transactions directly within messaging apps, connecting to existing CRMs, booking tools, and payment processors. 

Before integrating Claude, Vambe faced three technical barriers. Previous AI models would lose context mid-conversation, repeat information, or struggle with regional language variations—a significant problem when a customer in Mexico City uses different expressions than one in Buenos Aires, and the model needs to understand both fluently. Lead qualification required nuanced judgment that other models couldn't deliver: distinguishing a customer ready to buy from one just browsing, then responding appropriately. And as Vambe's platform evolved beyond chat to executing real business actions—updating CRMs, processing payments, scheduling appointments—they needed a model that could reliably follow complex instructions without errors.

Why Vambe selected Claude

Vambe's team tested multiple AI models during development. Claude consistently outperformed on the capabilities that mattered most for commercial automation: checking inventory databases, calculating pricing, booking appointments, updating CRMs, and processing payments.

”The main difference was function calling reliability,” says Diego Chahuán, Co-founder of Vambe. “Vambe's platform needs to execute real business actions. When an AI model misunderstands a function call or executes it incorrectly, it doesn't just break the conversation—it costs the customer money or damages their reputation. Claude got this right far more consistently than other models.” 

Claude also handled the nuances of Latin American Spanish effectively. The language varies significantly between Mexico, Colombia, Argentina, and Chile—different slang, different ways of asking the same question. Claude maintained conversation quality across regions without requiring separate models or extensive fine-tuning.

"Context retention made a practical difference,” Chahuán says. “Claude could pick up a conversation from three days ago and remember what was discussed, what prices were quoted, what objections came up. This meant fewer lost deals.”

Vambe automates sales conversations with Claude

Vambe had built their entire product vision around function calling—AI agents that could actually execute actions, not just respond to messages. But existing models simply weren't accurate enough. The team was experiencing error rates that meant customers couldn't trust the AI to handle real transactions without supervision.

"When Sonnet 3.5 launched and we saw it was significantly better at coding—where function calling is a critical skill—we were thrilled. From the very first moment we tried it, it felt like discovering fire," says Chahuán. Developers are now using Claude Opus 4.5 and Claude Code. 

Function calls that had previously failed or required multiple attempts suddenly worked reliably. The team could chain multiple API calls together, handle edge cases they had written off as impossible, and trust the AI to process payments or book appointments without human verification at every step.

Implementation moved quickly. Vambe had Claude running in production within two weeks. From decision to full deployment: three weeks.

Claude now powers Vambe's conversational AI platform across multiple industries. In the automotive sector, dealerships use Claude-powered agents to qualify leads and guide customers through vehicle selection, financing options, and test-drive scheduling—all within WhatsApp. In healthcare, clinics use Claude to manage appointment scheduling, reminders, and patient follow-ups, with some clinics reducing no-shows by up to 60%. In e-commerce and retail, Claude supports personalized product recommendations and real-time order assistance. In education, institutions rely on Claude-integrated agents to guide prospective students through program discovery, enrollment, and financial aid queries.

Aki KB, a Chilean storage rental company, saw response times drop from two to three hours to seconds. "One of our biggest fears was losing our personal touch—we've always been known for that," the Aki KB team explained. "That's why it was so important to find the right mix between technology and human service, and Vambe delivers that perfectly."

One pattern stands out: businesses name their AI assistants—Francisca, María, Pedro—and most customers don't realize they're speaking with AI. Some even show up at physical offices asking to speak with them. 

The outcome

The most significant improvement with Claude was reliable AI-to-AI handoffs. Previously, handoffs between agents failed 30% of the time. With Claude, reliability improved to over 95%, and the average number of agents per platform increased from 1.0 to 2.3.

This drove measurable business impact across Vambe's customer base: 40-60% higher conversion rates compared to previous sales approaches, 60% fewer no-shows in healthcare settings, and the ability to handle two to three times more customer volume with the same team size.

For Vambe's engineering team, debugging time dropped from 40% of their workload to 10%, freeing 120 hours per month for feature development. Customer onboarding went from seven days to three days.

"Claude turned our multi-agent architecture from a technical limitation into our main competitive advantage," says Chahuán. 

Vambe's immediate focus is Brazil, Latin America's largest market. Beyond geographic expansion, the company is exploring predictive customer insights, voice interactions for phone-based sales, and agentic workflows that coordinate both customer-facing and internal operations.

"We don't see Anthropic as just a model provider,” Chahuán says. “We’re making advanced AI accessible to businesses that have historically been left behind by technology. Latin America represents 650 million people and millions of businesses that have been underserved by enterprise software. With Anthropic, we're building that future now.” 

‍