Customer Service Automation
Intercom needed a flexible AI platform that could keep pace with rapidly changing models while handling large volumes of customer support interactions, complex workflow tasks, and new voice channels at production quality.
Reported impact
Intercom launched Fin just four months after GPT-4, replatformed the business around AI, migrated task automation onto GPT-4.1 in days, and reduced cost while increasing reliability and support quality across text and voice channels.
Intercom reorganized its product and engineering stack around AI so Fin could move from experimental chat to a durable service platform. The company launched Fin within months of GPT-4, then built a modular architecture that routes support work across chat, email, and voice while swapping in better models as they emerge. Fin handles the conversational support layer, Fin Tasks automates multi-step workflows such as refunds or account changes, and Fin Voice uses the Realtime API for phone-style interactions. GPT-4.1 now powers a growing share of Intercom's AI usage because evals showed it could complete key task flows with strong reliability, lower latency, and a 20% cost reduction versus GPT-4o. Structured offline tests, live A/B experiments, transcript-based benchmarks, and modality-specific voice evals let Intercom move from one model generation to the next in days instead of quarters without rewriting the surrounding system.
Customer interactions flow into Fin over chat, email, or voice. The platform routes conversational work to the best model path, invokes Fin Tasks when refunds or other multi-step actions are needed, evaluates behavior with transcript and A/B tests, and feeds connected conversation and ticket data through the MCP Server for downstream product and operations teams.
"GPT-4.1 powers a growing share of Intercom's AI usage, including key logic inside Fin Tasks."
"The architecture allows Intercom to route queries to the best model for the job and swap models without reengineering the underlying system."
"When GPT-4.1 dropped, we had eval results within 48 hours and a rollout plan right after"
"Within hours of GPT-3.5's release, the customer service software company began experimenting, and just four months later launched Fin"
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.
The public company page includes numberOfEmployees value 1030 and a PostalAddress with addressLocality San Francisco and addressCountry US.
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 