A reference for every trigger, condition and action you can use in your flows.
Starts the flow when a Live Chat message arrives.
How it works
💡 Great for an automatic welcome reply.
Listens for inbound chat-widget messages. Pair it with a 'Send message' node to reply instantly, or with Melia AI for a knowledge-base-grounded auto-reply.
Starts when the visitor clicks the chat icon.
How it works
💡 Show quick replies to guide them right away.
Fires when the visitor clicks the widget launcher. The ideal moment to present a choice menu (quick replies / buttons) and route the conversation.
Starts on the visitor's very first visit.
How it works
💡 Keep the message short and welcoming. It's the first touch.
Fires when the widget detects a first visit. Useful to proactively engage a new visitor. Requires an active widget session to deliver the message (anonymous pre-session delivery is an upcoming complement).
Starts when a returning visitor comes back.
How it works
💡 Offer to resume the conversation or a loyalty perk.
Fires for a returning visitor (already visited / had a conversation). Great to personalize the welcome or recover a cart.
Starts when the visitor opens a given page.
How it works
💡 Target /pricing to offer help at purchase time.
Matches the current URL against the pattern you enter (substring, case-insensitive). Leave empty to react on any page. Perfect for per-page contextual messages.
Starts when the visitor scrolls past a threshold.
How it works
💡 A 50-75% threshold targets engaged visitors.
The widget reports scroll depth in steps; the flow starts when the visitor passes the configured percentage. A good interest signal to engage at the right moment.
Starts when the visitor is about to leave the page.
How it works
💡 Offer a discount or help right before they leave.
Detects exit intent (the mouse leaves the viewport through the top). Lets you show a retention message at the last moment, before the visitor closes the tab.
Starts the flow when a WhatsApp Business message arrives.
How it works
💡 Beyond 24h, use an approved template via 'Send notification'.
Listens for inbound WhatsApp Business messages (Meta Cloud API). Free-form replies are only allowed within the 24h window after the contact's last message; outside it, only approved templates go through.
Starts the flow when a Messenger message arrives.
How it works
💡 Pair with a condition to route by message content.
Listens for inbound Facebook Messenger messages. Like WhatsApp, Messenger enforces messaging windows: out-of-window replies are restricted by Meta.
Starts the flow when an inbound email arrives.
How it works
💡 Filter by subject to route 'invoice' emails differently from 'support'.
Email has no channel: it scopes by reception address (the channel analog) plus optional sender (domain / exact / contains) and subject filters. All filters empty = reacts to every inbound email.
Starts the flow when an inbound SMS arrives.
How it works
💡 Leave the channel empty to react to all your SMS numbers.
This trigger listens for inbound SMS on the SMS channel. For each SMS received, the runtime looks for a flow whose entry is an SMS trigger and starts an execution. If you set a channel, only SMS received on that channel start the flow.
Continues here when another flow hands off.
How it works
💡 Split large flows into reusable sub-flows.
This trigger never fires on its own: another flow must target it via its 'Go to another flow' node. Parameters collected upstream are passed in, and on return the sub-flow's parameters are merged into the parent.
Starts when a conversation is marked resolved, on any channel.
How it works
💡 Use it to send a satisfaction survey or a thank-you follow-up.
Starts a flow the moment a conversation is marked resolved, whatever the channel. This is the server-side counterpart to the agent clicking Resolve: you don't need the visitor to send anything.
Typical uses: send a CSAT / satisfaction survey, a thank-you message, a follow-up offer, or notify another system that the ticket is closed.
It runs once per conversation. If a conversation is reopened and resolved again, the flow does not fire a second time, so you never spam the same contact.
Starts when a conversation sits unassigned longer than a delay.
How it works
💡 Use it to auto-route to a department or ping a manager when nobody picks up.
Starts a flow when a conversation has been left unassigned (no agent has taken it) for longer than a set delay (10 minutes by default, adjustable by an administrator).
This is a server-side, time-based trigger: a background sweep runs every couple of minutes and starts the flow for any open, unassigned conversation that has waited too long. It is not tied to the visitor sending a new message.
Typical uses: auto-route to a department or team, escalate to a manager, send an "we're on it" holding message, or raise an internal alert. It runs once per conversation, so a long wait won't trigger the flow repeatedly.
Branches on the value of a saved parameter.
How it works
💡 Capture the value first with 'Ask a question' or 'Form'.
Compares a flow parameter to each of your branches. The first match wins; otherwise the 'else' branch is taken. Parameters come from nodes that capture an answer.
Melia classifies the message into an intent and routes.
How it works
💡 Describe each intent well to guide classification. The ✨ button drafts the description from the title (billed as one AI response).
Melia (AI) analyzes the inbound message and classifies it into one of your intents, then takes the matching branch. Below the threshold, the 'else' branch is taken. No knowledge base is needed: classification uses the AI model configured by the platform.
Branches on the day and time.
How it works
💡 Route to an agent during opening hours, to a message otherwise.
Checks whether 'now' falls on a selected day and inside the time window, in the given timezone (IANA, e.g. Europe/Paris; empty = UTC). Use it for business hours.
Branches on whether an agent is online.
How it works
💡 Offline: offer a message form instead of an agent.
Looks at whether at least one agent is online for the organization. Lets you adapt the path: connect to an agent if someone is available, otherwise collect a message for follow-up.
Branches on whether the contact is in a group.
How it works
💡 Pair with 'Add tag' to segment your paths.
Tests whether the current contact belongs to the selected group (groups act as tags). Great to treat VIP contacts, newsletter subscribers, etc. differently.
Branches on the visitor's language.
How it works
💡 Serve a per-language path from a single flow.
Compares the contact's language (from profile / browser) to the comma-separated codes you list. Matching is by prefix, so 'fr' covers 'fr-FR' and 'fr-CA'.
Branches on the visitor's country.
How it works
💡 Adapt currency, language or offer by country.
Tests the contact's country code (ISO-2) against the provided list. Useful to offer region-specific content, currencies or delivery times.
Branches on new vs returning visitor.
How it works
💡 Welcome a newcomer differently from a regular.
Treats a visitor as returning if they already had a conversation (more than one ticket). Lets you adapt the welcome: an intro for a newcomer, a resume for a regular.
Branches on the visitor's device.
How it works
💡 Simplify the path on mobile (shorter messages).
Infers the device type from the visitor's latest connection User-Agent. Lets you tailor the experience (e.g. fewer steps, adapted formats) for mobile/tablet/desktop.
Branches on the last page viewed.
How it works
💡 Detect visitors coming from a specific product page.
Looks at the visitor's last page and checks it contains the entered substring. Great to contextualize based on the browsing journey.
Sends a sequence of messages to the contact.
How it works
💡 Space messages with a short delay for a natural pace.
Walks your messages and sends them one after another, honoring the delay between sends. {{...}} variables (contact name, etc.) are substituted at send time.
Asks a question and stores the answer.
How it works
💡 Reuse the parameter later via `{{parameter_name}}`.
Pauses the flow until the contact replies, validates against the chosen type (retries if invalid up to the retry count), then saves the value into a parameter reusable downstream.
Collects several fields in one form.
How it works
💡 More efficient than chaining several 'Ask a question'.
Shows a form in the Live Chat widget; on submit, each answer is saved as a parameter (the field key) then the flow continues. Great to collect name + email + need at once.
Offers choices and routes by the reply.
How it works
💡 In Live Chat, cards (image+title) are highly engaging.
Presents your options to the contact. In Live Chat, pick the rendering: buttons, quick replies (chips) or cards (image + subtitle). On SMS/Email, a numbered menu is sent. The reply takes the chosen branch.
Randomly splits across weighted branches.
How it works
💡 Compare two messages 50/50 to see which performs best.
On each pass, a branch is drawn at random, proportional to its weight. Lets you test several path variants and measure which converts best.
Pauses the flow for a duration.
How it works
💡 Wait a few minutes before a follow-up to avoid being intrusive.
Suspends execution and resumes automatically once the duration elapses, even for long delays (hours/days). Useful for spaced follow-ups or timed sequences.
Adds or removes the contact from a group (tag).
How it works
💡 Tag 'hot lead' to follow up on those contacts later.
Changes the contact's membership in the chosen group (groups act as tags for segmentation). Pair with the 'Has tag' condition for targeted paths.
Updates a contact field.
How it works
💡 Save a collected answer into the contact record.
Writes a value into a contact field. Email/phone fields append to the existing list; others replace the value. Useful to enrich the record as the conversation goes.
Marks the conversation resolved or blocked.
How it works
💡 Place it at the end after a closing message.
Changes the ticket status and, in Live Chat, tells the widget the conversation is closed. Use it at the end of the flow, usually after a final message.
Subscribes the contact to the org newsletter.
How it works
💡 Place it after an explicit contact consent.
Adds the contact to the organization's default newsletter group (created automatically if needed). Remember to collect consent before subscribing.
Assigns the conversation to an agent.
How it works
💡 Leave 'unassigned queue' so any agent can pick it up.
Assigns the ticket to the selected agent (and notifies them), or sends it to the shared unassigned queue every agent sees. For team routing, use 'Route to department' instead.
Routes the conversation to a department/team.
How it works
💡 Create your departments in Settings → Departments & teams.
Assigns the ticket to the chosen department (and optionally team). Per the group's method, an agent is picked automatically (round-robin, preferring online agents) and notified; in manual mode, the ticket waits in the department queue.
Hands off to another flow.
How it works
💡 Reuse common sequences (e.g. contact capture).
Launches the target flow and waits for it to finish. On return, the sub-flow's parameters are merged into the current flow. If the sub-flow fails, the failure branch (if set) is taken.
Calls an external API and routes on the response.
How it works
💡 Fetch an order status then personalize the reply.
Performs an HTTP call (with retries) to the service of your choice. The response can be stored into a parameter via a path (JSONPath). 2xx → success, otherwise → failure.
Notifies an agent or internal channel.
How it works
💡 Alert the team when an urgent case arrives.
Fires one or more outbound notifications to the chosen channels, with a delay between each. Email / WhatsApp Business sends use templates with parameter mapping.
Melia answers from a knowledge base.
How it works
💡 Wire 'handoff' to an agent for uncovered cases.
Melia generates a reply grounded in your knowledge base and posts it as the AI agent. Depending on the outcome (answer found, abstained, turn cap reached, error), the flow takes the matching branch.