AI BriefingImplementing High-Performance Customer Support Service
Key point
Decagon fully automates large-scale customer support processes by leveraging a variety of OpenAI models.
Details
Decagon provides large-scale customer support automation solutions for global companies such as Curology, Duolingo, and Notion. By combining a variety of OpenAI models—GPT-3.5, GPT-4, GPT-4o, and o1-mini—it implements agentic bots that manage the entire customer lifecycle, going beyond simple response generation.
The key is deploying the optimal model according to the nature of the task.
- GPT-3.5 (fine-tuned): Used for rewriting customer queries before they enter the RAG workflow, maximizing performance
- GPT-4: Handles sophisticated tasks such as complex decision-making and API requests
Through this approach, Decagon has achieved results in which 91% of a large customer's global customer support operations are handled without human intervention. It also maintains a flexible infrastructure that allows it to quickly evaluate and integrate new models as soon as they are released.
Going forward, Decagon aims to extend its AI agents' capabilities into the voice domain, achieving a high level of automation and accuracy in voice-based customer support as well.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.