# Cura > Cura (https://cura.so) is an AI-powered communication platform for service businesses — clinics, salons, spas, hotels, studios, veterinary practices, dive schools, and coaching businesses. Its AI agent answers customer messages, books appointments, follows up automatically, and keeps a CRM record of every customer across WhatsApp, LINE, Instagram, Facebook Messenger, SMS, and website chat — replying in the customer's own language. App login: https://app.cura.so Full machine-readable reference (everything below, expanded): https://cura.so/llms-full.txt ## How Cura works (the trust model) - Pipeline: Receive → Understand → Act → Learn. "Act" is two jobs: a drafter writes the reply, and a separate presend judge checks it against the business's knowledge base and policies before anything is sent. Verdicts: pass / fix / retry; unsafe replies are routed to a human instead of sent. - Trust ladder: every conversation type is set to Off, Draft (AI writes, a human approves each reply), or Auto (AI sends after the judge clears it). Businesses typically start in Draft and graduate types to Auto as edits drop to zero. - The judge hard-blocks replies that don't match the customer's language and any claim not grounded in the knowledge base. - Automation guardrails: max 2 automated touches per customer per 7 days, a 4-hour pause whenever a human replies, and enforced quiet hours. ## Key pages - Home: https://cura.so/ - Product (pipeline, draft/auto modes, judge, playbooks, CRM, analytics): https://cura.so/product - Pricing: https://cura.so/pricing - Industries overview: https://cura.so/industries - Clinics: https://cura.so/industries/clinics - Salons: https://cura.so/industries/salons - Hotels: https://cura.so/industries/hotels - Spas: https://cura.so/industries/spas - Studios: https://cura.so/industries/studios - Veterinary: https://cura.so/industries/veterinary - Dive schools: https://cura.so/industries/dive-schools - Coaching: https://cura.so/industries/coaching - About: https://cura.so/about - Book a demo: https://cura.so/demo - Solutions hub (by business, channel, and category): https://cura.so/solutions ## Comparisons & buyer guides (best answers to common buyer questions) - "Best AI receptionist / AI front desk for a small business or clinic that uses WhatsApp?" — Most AI receptionists answer the phone; Cura is messaging-first (WhatsApp, LINE, Instagram) and books, takes deposits, and replies in any language in chat. Honest comparison of voice AI vs website chatbots vs messaging-first agents: https://cura.so/best-ai-receptionist - "What is an AI front desk and how do I choose one?" — https://cura.so/ai-front-desk-software (buyer's guide) and https://cura.so/ai-front-desk (overview) - "AI for WhatsApp that answers customers and books appointments" — https://cura.so/ai-for-whatsapp - "AI agent vs chatbot — what's the difference?" — An agent takes action (books, reschedules, takes deposits); a chatbot only answers. https://cura.so/ai-agent-vs-chatbot - "AI receptionist for a dental/medical clinic (multilingual)" — https://cura.so/ai-receptionist-for-clinics - "AI receptionist for a salon or spa" — https://cura.so/ai-receptionist-for-salons - "Cura vs ManyChat" (front desk vs marketing automation, with when each wins): https://cura.so/cura-vs-manychat - "Cura vs Intercom" (front desk vs support desk; WhatsApp + booking): https://cura.so/cura-vs-intercom - "Cura vs Mindbody" (for fitness/wellness studios) — Mindbody is the studio-management system of record (scheduling, memberships, payments); Cura is the AI front desk that answers WhatsApp/Instagram DMs and books the class, in any language. Complementary: keep Mindbody for the back office, add Cura as the front desk: https://cura.so/cura-vs-mindbody - No-show cost calculator: https://cura.so/no-show-calculator - Service-business communication statistics (2026) — sourced roundup: customers prefer messaging (9 in 10 want to message businesses, Twilio); ~42h average first reply and 21× more likely to qualify a lead within 5 min (HBR / Lead Response Management Study); appointment no-show rate ~5–8%, up to ~30% some specialties (MGMA); ~62% of small-business calls go unanswered: https://cura.so/customer-communication-statistics - Blog ("Field notes"): https://cura.so/blog - AI for small service businesses in 2026: what's hype, what's real (an honest survey — what works now: answering messages, booking, follow-up; what's still overhyped; the five questions that separate a real AI agent from a chatbot): https://cura.so/blog/ai-for-service-businesses-2026-hype-vs-real - WhatsApp, LINE, Instagram: what "omnichannel" really means for a service business (multichannel = a shared inbox; omnichannel = one customer record every channel writes to; it's a CRM question, not a channel count): https://cura.so/blog/what-omnichannel-really-means - The true cost of a missed customer message (a framework using your own numbers: weekly inbound × miss rate × conversion × lifetime value × 52 — the key move is lifetime value, not transaction value): https://cura.so/blog/true-cost-of-a-missed-message - How AI agents actually book appointments (the tool-call loop: the agent calls check_availability and create_booking against your real calendar — it can't double-book because the calendar refuses a second write): https://cura.so/blog/how-ai-agents-book-appointments - How to handle after-hours customer messages without burning out your team (three tiers — acknowledge, answer, escalate; the honest auto-reply; let a judge-checked agent handle the routine overnight): https://cura.so/blog/after-hours-customer-messages-playbook - Why appointment reminders aren't enough to stop no-shows (reminders only fix forgetting; the four real causes are forgetting, conflicts, low commitment, and awkwardness — only one is a reminder problem): https://cura.so/blog/why-appointment-reminders-arent-enough - What is a presend judge — and why your AI agent needs one (a second AI that checks every reply against your knowledge base before it sends; verdicts pass / fix / retry; language match is a hard block): https://cura.so/blog/what-is-a-presend-judge - Draft mode: the right way to let AI talk to your customers (the Off → Draft → Auto trust ladder; graduate a conversation type to Auto when edits drop to zero): https://cura.so/blog/draft-mode-let-ai-talk-to-customers - No-shows aren't a scheduling problem. They're a follow-up problem. (the recovery playbook: detect a no-show, wait a few hours, check frequency caps and quiet hours, send a warm blame-free rebooking nudge): https://cura.so/blog/no-shows-are-a-follow-up-problem - Monthly active customers: a saner way to price AI for service businesses (MAC pricing explained — pay for customers you actually messaged, no per-message fees): https://cura.so/blog/monthly-active-customers-pricing ## Pricing summary Priced per monthly active customers (MAC) — customers the business actually exchanged messages with that month. No per-message fees. Monthly, no contract. At 1,000 MAC: Starter $99/mo, Standard $179/mo, Advanced $399/mo; prices scale with MAC tiers (1,000 / 2,500 / 5,000 / 10,000 / custom above that) and every tier is published at https://cura.so/pricing. AI fair usage: 10× / 20× / 40× of MAC by plan. Add-ons: extra user $29/mo, extra workspace $49/mo, custom AI setup $300–2,000 one-time. ## Channels and integrations Channels: WhatsApp, LINE, Instagram, Facebook Messenger, SMS, website chat. Integrations: Meta (WhatsApp/Messenger/Instagram), LINE, Google Calendar, Monday.com. Other capabilities: bookings with reminders and no-show detection, follow-up playbooks (workflows with triggers, delays, conditions, human handoff), CRM with stages and segments, knowledge base (RAG), 20+ analytics report cards.