Describe your bot, or build it by hand — either way it stays yours to read.

A visual editor where you compose a chatbot conversation by dragging nodes onto a canvas and wiring them together. Each node is one step — send a message, ask a question, branch on the reply, call an API, update a contact. The bot follows whichever path you authored, top to bottom. No code, no scripting language, and no LLM deciding replies at runtime — just a deterministic flow you can read and audit.

Yes — AI Flow Generation lets you type a plain-English description ("book appointments for my salon") and get a complete draft flow on the canvas: nodes, questions, and branches already wired. It's a starting point you review and edit, not an autonomous agent — you open each node, adjust the copy and logic, then publish. AI Flow Generation is included on Professional and Reseller plans (it's off on Free).

Yes — that's the whole point of the canvas. The only place "code" enters is when you configure an API Request node, and even there you're filling in a URL, headers, and a JSON body, not writing a function. If you can build a Zapier zap, you can build a WhatsMark flow. Or skip that entirely and let AI draft the first version for you.

Two ways. The Condition node is a clean two-way yes/no split on the last reply text, matched by contains, equals, starts-with, ends-with, or regex — and it requires both branches to be connected, so you can't accidentally build a dead end. For richer routing, the Logic Control node checks many rules in order (first match wins, with a mandatory fallback) and can branch on a saved variable, a contact custom field, or a value returned by an earlier API Request — not just the message text.

Yes — the API Request node makes an outbound HTTP call mid-flow (GET/POST/PUT/PATCH/DELETE) and passes variables you've captured — order ID, phone, replies — in the request. The Save Response node then maps fields out of the response into named variables you use downstream: show them in a message, branch on them with Logic Control, or pass them to the next call. Common patterns: fetch order status, look up a contact, check availability, verify a coupon, push a lead. Full protocol detail lives on the webhooks & API page.

Honestly: there isn't a single "escalate" node on the canvas today. The pattern most teams use is a fallback branch — when the flow hits its limit (user asks for a person, or an API call fails), it sends a "let me get someone to help" message and stops the automation, and the conversation is already sitting in your shared team inbox for an agent to pick up. Agents can also start a flow on a live conversation from the inbox. A dedicated in-canvas escalate-to-agent node is on our roadmap, not shipped yet — so we won't claim one that isn't there.

Different things, despite the shared word. Meta's WhatsApp Flows is a form widget that opens inside a chat — date pickers, dropdowns, multi-page forms — collects structured data, then closes. Our bot flow builder is the conversation engine — the bot that decides what to send, when to branch, when to call your API, what to do next. The two are complementary; WhatsMark ships the interactive WhatsApp Flows product as its own feature, and the bot flow builder is where you author the conversation around it.

Yes — variables. The Ask Question (and related Ask) nodes save the reply to a named variable. That variable is then available downstream: substitute it into a message, pass it to an API Request, or evaluate it with the Logic Control node. The Save Response node does the same for API responses — pulling specific fields out of a JSON reply into variables you can reuse.

A flow begins at its Trigger node. Message triggers fire on an exact keyword match, when a message contains a keyword, on a contact's first-ever message, or as a catch-all fallback when nothing else matches. Flows can also start from store events — Shopify and WooCommerce cart and order events are first-class trigger sources, so a flow can react to an abandoned cart or a new order, not just an inbound message.

Yes — starter flows like abandoned-cart recovery, appointment booking, and order tracking that you clone into your account and edit to match your backend and copy. They're starters, not the catalog — the real catalog is whatever you build on the canvas, and AI Flow Generation can draft a brand-new one from a sentence.

The builder is on every plan, with no add-on and no per-agent fee — the only thing that scales is how many flows you can save at once: 1 on Free, 15 on Professional, unlimited on Reseller. Compare that with AiSensy (a paid Flow Builder add-on after the trial), BotPenguin (capped bots on lower plans), or WATI's per-user laddering. Meta's per-conversation charges still apply at 0% markup, same as everywhere else on the platform.