Customer Service Automation
Decagon needed to automate complex customer support conversations at high volume while keeping latency low, preserving nuanced responses, and adapting to each client’s business logic and workflows.
Reported impact
Decagon enabled highly automated support for large clients, shortened onboarding for new deployments, and kept its platform current by quickly evaluating new OpenAI models for accuracy and efficiency gains.
Decagon built a configurable support-automation platform that combines multiple OpenAI models across the customer-service pipeline. Fine-tuned GPT-3.5 rewrites incoming customer queries before they enter retrieval-augmented generation workflows, while GPT-4 handles more complex decisions such as API-driven actions and intricate operations. The broader platform can also draw on GPT-4o, GPT-4 Turbo, and OpenAI o1-mini so each stage of the interaction uses the model best suited for its speed and reasoning needs. This model orchestration lets Decagon capture client-specific business logic, generate tailored software surfaces around each agent, and rapidly test new model releases through evaluations. Customers use the platform to run fast, mostly autonomous support operations at global scale, and Decagon can bring the core infrastructure online for a new customer in days while continuing to explore voice as the next automation frontier.
Incoming support questions are rewritten with fine-tuned GPT-3.5, passed through retrieval and business-logic workflows inside the Decagon platform, and then routed to GPT-4 or other OpenAI models for response generation or API actions. The resulting outputs are evaluated continuously and returned either directly to customers or used to expand automation coverage.
"Decagon fine-tuned GPT‑3.5 to rewrite customer queries before they enter retrieval-augmented generation (RAG) workflows."
"For new customers, our core infrastructure can be up and running in days"
"For one of our largest customers, we handle 91% of all their global support, without a human being involved"
"the company uses GPT‑4 for complex decision-making tasks"
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
OpenAI Sample Sep 22, 2026 Ringg Ringg needed to help enterprises automate large volumes of customer calls and chats with reliable tool use, low latency, multilingual quality, and unit economics that improved rather than worsened at scale. 4.8 average customer CSAT GPT-4.1 · GPT-5.6 Luna · GPT-5.6 Sol OpenAI Sample Aug 3, 2026 Circles Circles needed to turn rich telco signals such as usage, billing, network activity, and location into real-time support and monetization actions instead of reactive, siloed customer journeys. 22% increase in ARPU in Singapore from AI-driven personalization AI Concierge · CareX · Codex OpenAI Sample Jun 12, 2026 Preply Preply wanted to personalize language learning more deeply and reduce repetitive tutor work without replacing the human relationship that makes one-to-one teaching effective. 4.7/5 satisfaction rating Brand Voice GPT · ChatGPT Enterprise · Codex 