tawk.to
Helpdesk & ITSMConnect AI agents to tawk.to live chat. Agents create support tickets from inbound emails and form submissions, pull chat history for reporting, and list agents and properties to drive assignment and routing logic.
What This Integration Enables
tawk.to earned its ubiquity by making live chat free, which means the teams running it are usually small and stretched. Their real problem is not the chat, it is everything around the chat: the requests that arrive by email instead, the history locked in the dashboard, the routing that depends on whoever notices first. FlowRunner agents work that perimeter. They convert off-channel requests into proper tickets, pull chat history out for reporting, and use the agent and property rosters to drive assignment logic in the tools around tawk.to. - Convert inbound emails and form submissions into tawk.to support tickets automatically - Pull chat conversation history into spreadsheets or a warehouse for volume reporting - Look up ticket details to enrich and route follow-up actions in other tools - List agents and properties to build assignment and account-mapping logic - Sync ticket status and requester data into a CRM or notification channel One honest note: tawk.to's REST API is gated behind an approved API application, so this connector's operations should be verified against the private documentation tawk.to issues with your credentials before production use. That is not a reason to skip the automation; it is a reason to stand it up deliberately, verify each operation against your approved docs, then let the agents run. FlowRunner's [human-in-the-loop](/concepts/human-in-the-loop) gate covers the judgment calls either way, and the read-heavy shape of this connector means most of its value, reporting, routing, dedup checking, carries no write risk at all.
Without FlowRunner
With FlowRunner
Use Case Scenarios
The mailbox that stops leaking requests
Support email arrives via [Gmail](/integrations/gmail). The agent parses each message, checks List Tickets for an open case from the same requester, and calls Create Ticket for genuinely new issues with subject, body, and requester mapped. A [Slack](/integrations/slack) alert tells the team a case is waiting. The email channel stops being the place where requests go to be forgotten, and the chat team works one queue instead of two, with the dedup check standing between the mailbox and duplicate cases.
Chat volume you can finally see
Nightly, the agent pages through List Chats and appends each session's visitor, handling agent, and timestamps to [Google Sheets](/integrations/google-sheets). After a month, the sheet answers questions the dashboard never volunteered on its own: which hours actually need coverage, which agents carry the load, how often the same visitor returns with the same problem. Staffing conversations start from data instead of impressions, and the case for a second support hire writes itself or dissolves.
Routing built on the roster
The agent reads List Agents and List Properties to maintain a live map of who serves which widget. When a ticket's details come back from Get Ticket, the agent matches the requester's email domain against account ownership in the CRM and posts the case to the right owner's channel with the full context attached. High-priority tickets from named accounts skip the general queue entirely and land with their owner. tawk.to stays the system of record; FlowRunner supplies the routing intelligence the free tier was never going to build for you.
Human-in-Loop Highlight
Create Ticket is the single write in this connector, and it acts on a real visitor's behalf: it opens a case in their name that the support team will treat as the customer's own words, and follow-up lands in the visitor's inbox. The failure mode is not data loss, it is misattribution at scale. An agent bulk-converting a mailbox backlog can open duplicate cases for requests already resolved in chat, attach the wrong requester to a forwarded thread, or promote automated notifications into tickets a human then dutifully answers. So FlowRunner agents create tickets one at a time with a dedup check in front, and anything batch-shaped, a backlog import, a form-tool migration, gets staged for a support lead's review first: this many candidate tickets, these requesters, these subjects. A person prunes the list, and only then do the cases open. The team's queue stays trustworthy, which for a small support team is the whole game.
Agent Capabilities
6 actionsTickets
3- List Tickets Lists the configured property's support tickets, which capture offline messages and follow-up requests, with pagination. The dedup check before any ticket creation.
- Get Ticket Retrieves one ticket's full detail by ID: subject, message body, status, requester, and assignment. The context pull behind routing and enrichment.
- Create Ticket Opens a new ticket in the configured property on a visitor's behalf, converting an inbound email or form submission into a tracked case. The write this page's human gate exists for.
Chats
1- List Chats Lists past chat sessions with visitor, agent, and timestamp metadata, paginated. The export behind chat-volume reporting.
Agents
1- List Agents Lists the property's team members with names, emails, and roles. The roster behind assignment and reporting logic.
Properties
1- List Properties Lists the chat widgets available to your account, returning the property IDs that scope every other operation.
Frequently Asked Questions
What can FlowRunner do with tawk.to?
FlowRunner agents can run List Tickets, Get Ticket, and Create Ticket in tawk.to, plus 3 more actions.
Does connecting tawk.to to FlowRunner require OAuth?
No. tawk.to connects to FlowRunner with basic authentication (a username and password), no OAuth flow required.
Can tawk.to trigger a FlowRunner workflow automatically?
tawk.to doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.
Start building with tawk.to
$100 in credits. No card required. Connect in minutes.