Scaling Customer Service Without Scaling the Team: Lessons from Those Who've Done It with AI

The Synthesia Case: Surviving Exponential Growth

Synthesia's case is the most dramatic example: an increase from 40,000 to 316,000 monthly contacts (690% growth) that would traditionally require hiring approximately 150 new agents. Instead, the company implemented Intercom's Fin AI and achieved:

- 87% self-service resolution rate - 96% reduction in resolution time - 98.3% of queries self-resolved - 1,300 hours saved in 6 months

The critical lesson from Synthesia is: invest in the knowledge base before activating AI. The team spent weeks structuring, expanding, and validating their knowledge base before turning on the AI agent. This preparation was the difference between an AI that resolves effectively and one that gives wrong answers.

Another important takeaway is: gradual activation. Synthesia started with the simplest categories and progressively expanded to more complex ones, monitoring accuracy and satisfaction at each stage.

The Carrefour Case: Centralization with Personality

Carrefour demonstrates a third scenario: scaling service for multiple brands (Carrefour, Atacadão, Sam's Club) while maintaining distinct identity for each one:

- Response time from 4 hours to 40 minutes - 3 AI personas with distinct personalities - 80% remote operation with superior results - Intelligent escalation for complex cases

The unique lesson from Carrefour is: AI can be a brand identity tool, not just an efficiency one. The three personas (Carina, João, Samy) ensure that each brand communicates in its authentic tone, something that would be difficult to maintain consistently with a large team of human agents.

The 80% remote model also demonstrates that AI enables more flexible and decentralized operations — a competitive advantage for companies operating across large geographies.