Hello Careline
Start free

How it works

From first message to resolved outcome — one agent brain.

Customers reach you on phone, WhatsApp, webchat, Messenger, Telegram, Instagram, or SMS. Careline routes every conversation through the same AI agent, visual flow, and knowledge base — then hands off to humans when it matters.

Start free

Channels live

Every channel feeds one agent brain.

WhatsApp, Messenger, Instagram, Telegram, SMS, email, and your website chat all land on the same Careline AI agent — then flow nodes and knowledge take over.

  • → WhatsApp, Messenger, Instagram, Telegram, SMS, email, website chat
  • → Same AI agent + published flow for every ingress
  • → Messages pulse into one brain — then knowledge or human handoff

Same brain · two surfaces

Watch a short conversation on Messenger and your website widget.

Pricing, menus, and handoff — identical agent logic, different channel UI.

Architecture

Channel → Agent → Flow → Knowledge / LLM → Operator

Channels are doors into one agent brain. The published flow is the playbook. Knowledge and the LLM supply grounded answers. When confidence drops or someone asks for a person, Transfer opens the operator desk with the full transcript.

Channel receives the message or call

Twilio/SIP voice, WhatsApp Business, website plugin, Messenger, Telegram, Instagram DM, or SMS — each ingress hits Careline with identity and channel metadata.

Agent picks the published flow

The agent is configured with voice/language, LLM settings, knowledge bindings, and which flow version is live. One agent can serve many channels.

Flow executes node by node

Menus, messages, forms, knowledge queries, API/tools, OTP, shop/payment, conditions, and transfer nodes run in order — the same graph for chat buttons or voice prompts.

AI answers from grounded knowledge

Knowledge nodes retrieve FAQs/docs; the LLM stays inside your temperature, tools, and guardrails. If confidence drops or the user asks for a person, Transfer fires.

Operator continues with context

The desk shows channel, transcript, AI activity, assignee, and tools (send product, payment QR) so humans never start from zero.

How it works

Animated Careline screens.

Flow builder, Cara, operator desk, campaigns, and business — mocks shaped like the real product.

Channels & agents

How channels talk to your AI agent

WhatsApp, voice (Twilio/SIP), website chat, Messenger, Telegram, Instagram DM, SMS, and email all open conversations against the same agent. Ingress normalizes the event, then the agent’s live flow version runs — channel UX differs; the brain does not.

  • → Ingress normalizes messages/calls into one conversation thread
  • → Agent settings pick model, voice, language, and which flow is live
  • → Buttons vs spoken menus map to the same graph nodes
  • → Variables (language, booking IDs, OTP state) persist across turns
  • → Many channels can attach to one agent without duplicating logic
  • → Email threads share the same knowledge, Transfer, and campaign logic

Flows & Cara

Visual flows — built by you or drafted by Cara

Drag Start, Menu / buttons, Message, Knowledge, Transfer, End — loops back to the main menu are first-class. Or open Cara: “Create a hotel receptionist from the template bank.” Cara interprets the brief, selects skills, seeds the canvas, binds a knowledge base, and verifies draft quality. Improve with Cara anytime; Test before publish so every attached channel picks up the same graph.

  • → Improve with Cara on any open flow (Auto / Create / Improve / Review)
  • → Knowledge nodes query your FAQ / docs mid-conversation
  • → Main-menu loops and human Transfer are first-class edges
  • → Publish once — voice, WhatsApp, and webchat all follow

AI agents & LLM

Configure intelligence per agent — not a black box

Each agent owns provider/model, voice, temperature, tools, and knowledge bindings. Knowledge nodes retrieve FAQs/docs so the LLM stays grounded; guardrails block unsafe topics. Tools call availability, booking, or payment APIs. When confidence drops or the visitor asks for a person, Transfer fires — the AI never pretends to be human after handoff.

  • → Per-agent LLM + voice configuration
  • → Knowledge base binding for grounded answers
  • → Tools/APIs for availability, booking, payments
  • → Transfer when automation should stop
  • → Same agent can greet on voice and continue on WhatsApp

Knowledge & training

How your organization trains the Careline chatbot

Upload FAQs, PDFs, and policy docs into an org knowledge base. Careline chunks and indexes them for retrieval. Bind that base to an agent and drop Knowledge nodes in your flow — the LLM answers from your content, not the open web. Re-upload or edit anytime; Test before publish so every channel stays on-brand.

  • → Create an org knowledge base (FAQ paste, PDF, Word, Markdown)
  • → Careline indexes content for grounded retrieval
  • → Bind the base to agents and Knowledge flow nodes
  • → Ask test questions, tune guardrails, then publish
  • → Same trained brain on WhatsApp, voice, webchat, and campaigns

Operator desk

Humans join mid-conversation — with the full story

When the flow hits Transfer, the desk shows Waiting / Mine / All, channel badges, live status, and AI activity (“per-node flow activity”). Operators assign, reply, send a shop product, or push a BanglaQR payment — without restarting the journey. Handoff messages keep the visitor informed while a teammate takes over.

  • → Unified inbox: Web, WhatsApp, Telegram, Messenger, Instagram, Email
  • → AI activity panel shows bot progress on the flow
  • → Handoff copy keeps the visitor informed
  • → Shop “Send with Buy” + payment QR in the same thread

Campaign automation

Outbound that reuses your inbound brain

Campaigns schedule voice, WhatsApp, SMS, or email outreach — renewals, appointment reminders, cart recovery, clinic follow-ups, and nurture digests. Recipients who reply enter the same published agent flows and can escalate to the operator desk. Choose audience, channel mix, and pacing; quiet hours and rate limits keep outreach compliant.

  • → Select audience, channel mix (incl. email), and send pacing
  • → Attach the live agent + flow used for replies
  • → Track delivered / opened / answered / converted
  • → Failed or “talk to person” paths open desk tickets
  • → Same knowledge and Transfer rules as inbound

Business modules

Vertical ops beside the conversation stack

Booking, Shop, Hotel, and Clinic modules manage appointments, patients, doctors, services, and branches in the same org as agents and campaigns. Flows create or update those records via tools — so an AI booking on WhatsApp lands on the clinic appointments board your staff already uses.

  • → Clinic: appointments, assignments, branches, patients, doctors, services
  • → Search, status filters, cancel — same UI your ops team expects
  • → Same org as agents, campaigns, and channels

Careline Booking

How organizations schedule with Careline

Publish services and availability, then share a booking link or embed the widget on your site. Guests pick a day and time; Careline confirms the appointment. Pair with flows and campaigns so WhatsApp, email, or voice reminders fire automatically — an alternative to standalone schedulers like Calendly, connected to your AI agents.

  • → Create services & offerings (duration, resources, timezone)
  • → Share a public page or embed Careline Booking
  • → Guest selects date → time → details
  • → Optional: agent/campaign reminders on WhatsApp, email, SMS
  • → Same org as inbox, clinic modules, and knowledge

Ready to see it on your channels?

Spin up an agent, attach a flow, and start answering — or launch an outbound campaign the same afternoon.

Start free Book a 15-min demo Contact
Book a 15-min demo Start free