JivoChat

Put a bot in front of your JivoChat live chat. Agents receive visitor messages, agent joins, and chat-closed events in real time, reply with text, Markdown, or interactive button menus, and invite a human agent when the conversation needs one.

4 triggers 4 actions Custom auth available

JivoChat website ↗ Platform Documentation ↗ Capability data verified 2026-07-27

On Client Message fires as a visitor types their first question into the chat widget

Send Buttons Message answers instantly with a three-option menu: Sales, Support, Billing

The tapped button arrives back as a new On Client Message, and the agent drafts an answer from the docs and the visitor's account context

Send Markdown Message delivers the answer with formatted links to the relevant guide

The visitor's next message signals frustration, a pricing negotiation, or an explicit ask for a person

Invite Agent transfers the conversation to a live human agent with the full exchange intact

The support agent joins, confirmed by On Agent Joined, reads the transcript, and takes the conversation from there

When On Chat Closed fires, the transcript and outcome log to the helpdesk and the tracking sheet

What This Integration Enables

Live chat is where your highest-intent visitors are, and a bot's job there is triage, not impersonation. JivoChat's Bot API is built on that premise: it gives the bot real conversational tools, text, Markdown, quick-reply buttons, and then it gives it Invite Agent, a native event whose entire purpose is handing the conversation to a person. That makes JivoChat one of the few channels where FlowRunner's [human-in-the-loop](/concepts/human-in-the-loop) discipline is not bolted on but built into the protocol itself. - Receive visitor messages, agent joins, agent-unavailable events, and chat closures as real-time triggers - Answer common questions immediately, with formatting and links where the channel supports them - Route visitors through button menus instead of making them guess what to type - Escalate to a live agent the moment the conversation calls for one, with context preserved - Turn every closed chat into downstream work: tickets, CRM records, transcript logs

Without FlowRunner

The widget is a doorbell Live chat pings a shared inbox, someone answers when they can, and the visitor with a credit card open has left

Bot or human, pick one Either every chat burns agent time on password resets, or a chatbot stonewalls the customer who needed a person

Chats evaporate The conversation ends and nothing of it reaches the CRM, the helpdesk, or anyone who might have followed up

With FlowRunner

First response in seconds Every visitor gets an immediate, useful reply, a menu, an answer, an honest handoff, on the channel they wrote in on

Escalation as a built-in event The handoff to a person is a first-class action, fired the moment the conversation exceeds the bot's brief

Every chat leaves a record Closed conversations become tickets, CRM contacts, and logged transcripts without anyone copying anything

Use Case Scenarios

The front door that qualifies while it helps

A visitor on the pricing page opens the widget and asks about plan limits. On Client Message fires, the agent answers with Send Markdown Message, linking the relevant comparison, then follows with Send Buttons Message: "Want a walkthrough?" with Yes and No options. On Yes, the agent creates the visitor as a contact in [HubSpot](/integrations/hubspot) and fires Invite Agent so a salesperson joins while the intent is hot. The bot did the triage; the human arrived for the part that closes.

The after-hours chat that still captures the lead

At 11pm, On Agent Unavailable fires as a visitor asks about enterprise onboarding. The agent replies with Send Markdown Message: an honest away notice with the docs link and a promise of follow-up, then asks for an email with Send Text Message. The reply arrives as a new On Client Message, the agent writes the contact and conversation summary into HubSpot as an opportunity, and the morning team starts the day with a warm lead instead of an empty transcript.

The chat that becomes a ticket the moment it ends

A support conversation runs its course: bot answers, an escalation, a resolution from the live agent. When On Chat Closed fires, the agent opens a follow-up ticket in [Zendesk](/integrations/zendesk) carrying the transcript, and appends a row to [Google Sheets](/integrations/google-sheets) with the timestamp, topic, and whether a human was invited. Support leadership reads escalation patterns from the sheet weekly and tunes the bot's brief accordingly.

Human-in-Loop Highlight

Every Send Text Message reaches a real customer in a live conversation, and a bot answer cannot be unsent: a wrong refund promise or a tone-deaf reply to a frustrated visitor does its damage in the second it renders. That is why the decisive operation in this connector is Invite Agent, and why FlowRunner treats firing it as a duty rather than a failure. The bot's brief is bounded and explicit: answer the documented, route the ambiguous, and the moment a conversation involves money, anger, or an explicit request for a person, transfer it with the transcript intact. The visitor never re-explains, the agent never walks in blind, and the bot never negotiates a discount it had no authority to offer.

Agent processes routinely

Detects exception requiring judgment

Clear match Continues automatically

Ambiguous Routes to human via preferred channel

Human decides

Agent resumes with decision

Agent Capabilities

4 actions

Messaging

Send Text Message Sends a plain-text bot reply into an active conversation as a BOT_MESSAGE event, targeted by the chat_id captured from an inbound trigger. The core reply action; the message reaches the visitor on whichever channel they contacted you through.
Send Markdown Message Sends a Markdown-formatted bot reply, with bold, italics, links, and lists rendered on channels that support them and a fallback text shown on channels that do not. The action for answers that need structure and links.
Send Buttons Message Sends an interactive message with up to three quick-reply buttons. A tapped button arrives back as a new On Client Message event, which makes menus and yes/no prompts reliable instead of hopeful.
Invite Agent Transfers the conversation from the bot to a live human agent by sending an INVITE_AGENT event, using the client_id and chat_id from the inbound trigger. The escalation primitive this page's human gate is built on.

Triggers

4 triggers

Event Triggers

On Client Message Fires when a visitor sends a message in the chat, carrying the message text plus the chat_id and client_id every reply and escalation needs.
On Agent Joined Fires when a live agent joins the conversation. The signal that the handoff succeeded and the bot should stand down.
On Agent Unavailable Fires when no live agent is available to take the conversation. The cue for away messages, contact capture, and next-morning follow-up.
On Chat Closed Fires when the conversation ends. The moment to log transcripts, open tickets, update the CRM, and trigger satisfaction surveys.

Frequently Asked Questions

What can FlowRunner do with JivoChat?

FlowRunner agents can run Send Text Message, Send Markdown Message, and Send Buttons Message in JivoChat, plus 1 more action.

Does connecting JivoChat to FlowRunner require OAuth?

JivoChat uses a custom authentication method to connect to FlowRunner.

Can JivoChat trigger a FlowRunner workflow automatically?

Yes. JivoChat supports 4 triggers that can start a FlowRunner workflow automatically.

Start building with JivoChat

$100 in credits. No card required. Connect in minutes.

Start Building Free Talk to us