Also available
Add Fin to your app with the Fin Agent API
Embed Fin directly into your product with full control over the UI. No model integration or agent building required.
View documentation
introducing the
Fin API Platform
Build world-class customer experiences, agents, and products with the best-performing models for customer service.Leading brands deliver world-class service with the world's best models
Fin APIs
Explore our APIs
The Fin API platform includes APIs for answer generation and knowledge retrieval, available as a set or independently.
- [ 1 ]
Fin Apex API
Access the generative model at the core of Fin. Fin Apex is the best-performing customer service LLM, outperforming GPT-5.4 and Opus 4.5 on resolution rate, hallucination, and speed.
Key Stats- 1.12.8% higher resolution rate than the latest frontier models
- 1.265% reduction in hallucinations vs Sonnet 4.0
- 1.30.6s faster time to first token than nearest competitor
- [ 2 ]
Fin RAG API
Access Fin's unified retrieval-augmented generation pipeline over API, including Fin Apex 1.0, Fin Retrieval, and Fin Reranker models.
Key Stats- 2.171% average resolution rate
- 2.284% average resolution rate at the 90th percentile
- [ 3 ]
Fin Retrieval API
A custom retrieval model optimised for customer service use cases, with a carefully optimised tradeoff between efficiency and quality, proven at the largest production scales.
Key Stats- 3.1+20 pp. precision vs Voyage-Large-3
- 3.2Custom vector retrieval optimised on apps with 10s of millions of items
- [ 4 ]
Fin Reranker API
Use the heavy reranker powering Fin to sort candidate items by relevance, optimised for your LLMs context window.
Key Stats- 4.1+16.7% NDCG vs Cohere v 3.5, for Fin reranking
- 4.2Battle-tested heuristic and custom LLM powered context sensitive reranking
Use Cases
Use cases for the Fin API platform
Custom agents
01Build your own custom agents powered by the best models for customer service. Use Fin's retrieval, reranking, and generative models to enhance your existing agents or build new ones from scratch.
Vertical products
02Build vertical-specific agents for industries like hospitality, healthcare, and logistics, powered by the best-performing models for customer service.
Research
Unsupervised Learning Meets Generative AIWe designed a topic modelling system to improve the Fin AI agent by detecting...
How We Built a World-Class RerankerWe built our own reranker that outperforms Cohere Rerank v3.5, an industry-leading...
Using LLMs as a Reranker for RAGGood answers start with good context. Our AI agents use retrieval-augmented generation...
Finetuning Retrieval for FinAt Intercom, we’ve built Fin, an AI-powered support bot designed to understand users’...
Building out Intercom’s AI infraWe’ll discuss how we built tools to make our scientists productive at training modern AI...Building a Better Language Detection Model for FinWhen users ask Fin some question, they expect Fin to respond in the same language in which...Think Fast: Reasoning at 3ms a TokenWe step through the optimisation process required to make an Open Source reasoning model...
Research from our AI Group
Insights and blogs from our 60-person AI group building Fin at Intercom.
Latest Research
FAQs
FAQs
What is the Fin API Platform?
The Fin API Platform gives developers direct access to the specialized models that power Fin, including the Fin Apex generative model, the Fin Retrieval model, and the Fin Reranker model. These are available individually or as a complete RAG pipeline, enabling teams to build custom agents and products with the same AI that powers Fin.
What APIs are available through the Fin API Platform?
The platform offers four APIs: Fin Apex API (the generative model at the core of Fin), Fin RAG API (the full retrieval-augmented generation pipeline), Fin Retrieval API (a custom model optimized for customer service search), and Fin Reranker API (for ranking candidate content by relevance). Each is available independently or as part of the full pipeline.
How does Fin Apex perform compared to general-purpose LLMs like GPT or Claude?
Per Fin's published benchmarks, Fin Apex 1.0 delivers 12.8% higher resolution rates than the latest frontier models, a 65% reduction in hallucinations compared to Sonnet 4.0, and is 0.6 seconds faster to first token than its nearest competitor. It is purpose-built for customer service and grounded in your knowledge base rather than general training data.
What can I build with the Fin API Platform?
The platform is designed for two primary use cases: custom support agents built on Fin's retrieval, reranking, and generative models, and vertical products for industries like hospitality, healthcare, and logistics. The Fin Agent API also lets you embed Fin directly in your product with full control over the UI.
How do I get access to the Fin API Platform?
Access to the Fin API Platform starts at $250,000 in annual platform spend. You can request access on the fin.ai/api-platform page and the team will follow up based on your use case.