Fini (usefini.com) is an AI agent for customer support that resolves tickets across chat, email and voice in 130+ languages, takes actions in your systems, and maintains its own knowledge base. The answers below are short by design; each one links to the page that covers the topic in depth.

​
About Fini

Fini is an AI agent for customer support built for regulated industries such as fintech, banking, insurance and healthcare. It answers from approved knowledge, takes actions in your systems through deterministic workflows, keeps its knowledge base up to date, and escalates to your team with full context when it should. See What is Fini.
Fini is made by Fini Technologies Inc., founded in 2022 and backed by Y Combinator. Fini serves 100+ customers and handles 3M+ resolutions a month. For general questions, contact founders@usefini.com.
No. Fini runs on top of the helpdesk you already use, including Zendesk, Intercom, Front, HubSpot, Salesforce, Gorgias, LiveChat and Microsoft 365. Your tickets, queues and human team stay where they are. See Fini on top of your helpdesk.
A knowledge agent is live on day 1, agentic workflows by day 14, and full autonomy by day 30. Each milestone maps to concrete product steps, from connecting Sources to building Rulebook workflows and Test Suite checks. See Rollout timeline.
Fini supports 130+ languages across chat, email and voice. You control language at ingestion (the Language training preference in Sources), in knowledge scoping (language folders and Attribute filters in Articles), and in the agent’s prompts. See Supported languages.

​
How Fini works

Each message goes through Reply Rules, a Planning step, customer Attributes, then either a deterministic Rulebook workflow or an answer from approved Articles, and finally Guardrails before delivery. Every step is recorded in the AI Steps trace. See How Fini works.
From Articles, the curated knowledge graph that is the single source of truth for each agent. Content enters through Sources (help center, files, Notion, Google Drive, Confluence, Zendesk Help Center) and reaches Articles only after the review and publish workflow. See the Knowledge overview.
Fini’s agent reads whole approved articles rather than chunks pulled from an embeddings index, so answers are grounded in complete, reviewed content. See How Fini works and What is RAGless.
Yes. You connect Actions to your APIs and call them from Tool nodes in Rulebook behavior trees, which run deterministically: the same inputs produce the same path every time. See the Refunds use case.
A background AI pass detects knowledge gaps and conflicts in live conversations and drafts fixes, which always land in Review and never publish without human approval. Magic Articles and Generate Knowledge in Inbox turn raw content or resolved conversations into articles; their Status After Generation setting decides whether a draft goes to Review or straight to Published.
When a Planning escalation trigger matches, a Rulebook branch ends in escalation, a Guardrail can’t produce a safe reply, or the customer asks for a person. The conversation reaches your team with the transcript, knowledge used and reasoning trace attached. See Escalation and handoff.
Yes. Every Fini reply in Inbox has an AI Steps trace showing planning, attributes loaded, the rule that ran, how the answer was composed, the tags applied and the guardrail verdicts.
Yes. Start with an Internal Comment rule in Reply Rules so the agent posts internal notes only (shadow mode), and build regression checks in Test Suite before switching to direct reply.

​
Channels

Chat (the Fini widget on web, in your app and on mobile, plus helpdesk chat), email (helpdesk email, Microsoft 365 shared mailboxes or native Fini email), and voice. Fini also runs in a hosted Help Center, in Slack, and over the API. See the Channel overview.
Yes. Fini runs as a voice agent on inbound phone support, configured from Voice in the agent sidebar, and calls appear in Inbox with the same trace as chat and email. Phone-number provisioning is coordinated with your Fini contact. See Fini on voice.

​
Performance and accuracy

Fini’s stated benchmark across deployments is up to 90% resolution at 99% accuracy; it is not a guarantee for every account, and results depend on your knowledge and workflows. See How accuracy is measured.
To confirm (internal, remove before publish): www.usefini.com/trust-metrics says “resolution accuracy (95%)”, which conflicts with the 99% figure. Which one should this answer state?
In Fini Analytics, AI resolution rate counts only conversations with status Resolved by AI, while Deflection rate is 100% minus Human escalation rate, so it also counts conversations still waiting on the customer. Fini treats resolution rate as the metric that matters. See Resolution vs deflection.
Fini’s published figures are AI CSAT 4.9 versus human CSAT 4.7. You can track CSAT and sentiment for your own agent in Analytics.

​
Security and data

Fini is SOC 2 Type II and ISO 27001 certified and supports GDPR and CCPA. Policies, subprocessors and the pen-test report are on the Trust Center. See the Security overview.
Yes. Fini is HIPAA-compliant, with a BAA available for healthcare deployments. See HIPAA.
No. Customer data is never used to train foundation models. Data is encrypted with AES-256 at rest and TLS 1.2+ in transit, with US or EU data residency. See Data handling.

​
Pricing

Fini is priced per resolution: you pay when Fini resolves an issue end to end with no human handover, with no per-seat fees. See How pricing works and current rates at usefini.com/pricing.
A resolution is a customer issue Fini solves end to end with no human handover. See What counts as a resolution for edge cases.
Yes. Fini offers a 90-Day Money-Back Guarantee: 90% resolution in 90 days or you pay $0. Eligibility applies (new enterprise customers with 1M+ annual tickets, among other conditions). See Money-back guarantee and the guarantee terms.

What is Fini

The official explainer.

How Fini works

The lifecycle of one conversation.

Rollout timeline

Day 1, 14 and 30.

Security overview

Compliance, data handling and the Trust Center.