Intercom's AI customer service agent that resolves support queries across chat, email, voice, and social.






84 / 100
Overall Trust Score
Not disclosed
Fin's AI Engine includes proprietary models like 'fin-cx-retrieval' and 'fin-cx-reranker' that are purpose-built for customer service and custom-trained on real support interactions. These models are stated to outperform generic LLM approaches in retrieval accuracy. The Fin AI Engine is a patented, six-layer architecture that refines queries, retrieves relevant content using 'fin-cx-retrieval', re...
$49/mo
50 outcomes included; additional outcomes at $0.99 each. Sales qualifications are $9.99 each.
$99/mo
Includes 2,000 free Operator credits per month, then $0.01 per credit. Tiered pricing for conversations applies.
Custom
Custom pricing for Fin Voice is available. Contracted Pro plans include a credit allocation based on conversation volume. Fin AI can be bundled with I...
Fin's moat is primarily built on its proprietary, purpose-built AI models, such as Apex and fin-cx-retrieval/reranker, which are custom-trained on customer experience data and deliver superior performance metrics like high resolution rates and reduced hallucinations. The recent acquisition by Salesforce significantly bolsters its distribution, providing a substantial competitive advantage. While deep integration creates moderate switching costs, the lack of explicit proprietary data or network effects prevents a truly dominant position.
Intercom demonstrates moderate openness and transparency, particularly concerning its AI agent, Fin. The company provides extensive security and compliance documentation through its Trust Center, including SOC 2, ISO 27001, ISO 42001, GDPR, HIPAA, and AIUC-1 certification for Fin, which involves independent adversarial testing for safety and reliability. Intercom also emphasizes transparency in AI agent disclosure, allowing customers to configure how Fin is presented to end-users. They publish research and insights related to AI and contribute open-source SDKs and API descriptions. However, Intercom does not open-source the weights of its core AI models nor does it publish traditional transparency reports detailing government data requests or content moderation actions.
Intercom introduced Fin, an AI chatbot powered by large language models, initially using OpenAI's GPT-4.
Intercom reported that Fin could resolve nearly 50% of incoming queries automatically.
Intercom announced over 20 new features and capabilities for Fin, making it multilingual, enhancing answer quality, and providing deeper reporting.
Intercom launched Fin 2, which uses Anthropic's Claude model and includes the ability to perform actions for customers in addition to answering questions.
Intercom announced a broader focus for Fin, aiming for it to develop memory, accumulate deep knowledge of businesses, and interoperate with different tools and systems, moving towards a "Customer Agent future".
Intercom announced 12 major updates to Fin's Procedures and Simulations capabilities, enabling Fin to handle complex queries with improved deterministic controls and agentic behavior.
Intercom introduced Fin Apex 1.0, a post-trained language model for customer support resolution tasks, which became the core answering model for Fin.
Intercom raised $250 million in venture debt to fund continued development of its AI agent platform.
Intercom was renamed as Fin, signaling a complete pivot around its AI agent business.
Salesforce completed its acquisition of Fin.
426K monthly visits
medium tierFin positions itself as a leader in the new category of AI Customer Agents, delivering perfect customer experiences across support, sales, and commerce, and was recently acquired by Salesforce to boost its AI strategy.