FloGPT

You create a workspace, set your organization name, domain and timezone, and then give the AI something to know. In Knowledge you create a knowledge base and fill it with the Web Crawler (paste your site URL and start a crawl) or by uploading PDFs and documents. FloGPT chunks and indexes the content, keeps it fresh with scheduled recrawls, and returns answers with citations back to the source pages.

Next you deploy an agent. For chat, you create a widget, set the allowed domains, pick the knowledge base, and paste one script tag into your site (a React package is also offered). For voice, you attach a phone number and design the call flow in the visual Call Pipeline editor: greet, listen, look something up, call a tool, book or send an SMS, then end or transfer. Templates and vertical packs for dental, real estate, retail and support give you a starting flow.

From there the agent handles conversations around the clock. When it cannot resolve something, or the customer asks for a person, a ticket opens in the CRM with the transcript, contact identity and channel context. Auto-assign rules route it to a queue or agent, SLA timers start, and your team replies in the same thread while the AI drafts, summarizes or translates on request. Outbound calls are wired up separately through triggers such as webhooks, Shopify events, time-relative schedules or uploaded CSV lists, all running on the same pipeline.

• •

  Chat Widget - A brandable web chat installed with one script tag, with custom colors, avatars, launcher styles, file uploads, rich cards and escalation rules. Responses stream token by token, and the same agent can also answer on WhatsApp, Instagram and other channels.

• •

  AI Voice Agents and Call Pipeline - Voice agents answer and place calls, look up data mid-call (Shopify, your database, calendars, the knowledge base), book appointments with two-way Google or Microsoft calendar sync, send SMS during the call, and warm-transfer to a human with context. Flows are built visually with branching, tool calls, versioning and rollback.

• •

  Knowledge Base - Multi-source ingestion from websites, sitemaps, PDFs, DOCX files and FAQ sheets, with automatic chunking and embeddings, scheduled crawls and delta updates. Answers are grounded in your content and cite their source URLs.

• •

  Unified CRM Inbox - Web chat, voice calls, email, SMS, WhatsApp, Instagram and Facebook land in one ticket queue. Identity resolution merges an email address, phone number, WhatsApp number, Instagram handle and Shopify customer ID into one contact record, and tickets move through statuses with priority and SLA tracking.

• •

  Workflow Builder - A drag-to-connect canvas with LLM, knowledge base search, condition, HTTP call and tool call nodes, a realtime test preview, reusable sub-flows, multi-agent hand-offs, and A/B traffic splitting between published versions. The same flow can run in chat or on a phone line.

• •

  Outbound Triggers and Shopify Support - Six trigger sources (webhook, Shopify event, time-relative, internal event, data collection watch and scheduled list) fire personalized outbound calls with retry and backoff, quiet hours, opt-out lists, concurrency caps and SMS fallback. For Shopify stores the agent can track orders, cancel unfulfilled orders, recover abandoned carts and open a human ticket for refund requests rather than promising money on the call.

FloGPT fits small and mid-sized businesses that take a meaningful share of their customer contact by phone and want one AI agent covering both the website and the phone line. The use cases it documents include dental and healthcare practices booking and confirming appointments, real estate and auto dealers qualifying inbound leads, home services, hospitality and restaurants handling after-hours calls, and insurance, legal, logistics and education teams answering repetitive questions.

Shopify merchants are a specific focus: the store integration lets the agent answer "where is my order" calls and chats, call shoppers after an abandoned checkout, and recommend in-stock products from the live catalog. Support teams that currently run a chatbot, a help desk and a phone system side by side will find the unified inbox and identity resolution the main reason to look at it, since escalations arrive with the transcript attached instead of being retyped.

Businesses serving India get DLT-compliant SMS and calling through Jio and MSG91, voice support for Hindi, Tamil, Telugu, Kannada, Malayalam, Marathi, Bengali, Gujarati and Punjabi, and INR billing through Razorpay. Teams that want to test the chat side first can start on the Free plan, which covers 100 messages a month; voice agents and outbound triggers need a paid plan because they carry telephony costs from the moment they are enabled.