Typebot

Connect AI agents to Typebot, the open source conversational form and chatbot builder. Agents list typebots, read submissions and their execution logs, and drive live chat sessions programmatically on Typebot Cloud or self-hosted instances.

7 actions API key available

Typebot website ↗ Platform Documentation ↗ Capability data verified 2026-07-28

A scheduled sweep runs every fifteen minutes against the lead qualification typebot

List Results pulls the submissions collected since the last cursor position

Get Result retrieves each new submission's full answers and captured variables

The agent scores each respondent on the qualification answers: company size, budget question, timeline

Qualified respondents are upserted into the CRM with their answers mapped to contact fields

High-intent submissions post to the sales channel with the full conversation captured

The sales owner reads what the respondent actually said, in their own words, and decides whether this is a demo invite or a nurture sequence

What This Integration Enables

Typebot's open source pedigree shows in its API: everything the builder captures, the API exposes, including the execution logs most form tools keep to themselves. That transparency is what makes it automatable, and it holds whether the bot runs on Typebot Cloud or a self-hosted instance behind your own domain. FlowRunner agents sweep submissions into the systems where work happens, read per-session logs to catch the integration failures that silently eat respondents, and drive live chat sessions programmatically, turn by turn, against the published bot. - Sweep results on a schedule and sync answers and captured variables to CRMs, sheets, and databases - Retrieve individual submissions with every answer and variable from the session - Read execution logs per submission to detect webhook failures and integration errors while they are still fixable - Start and continue chat sessions programmatically, pre-filling variables to personalize or shorten the flow - Verify a published bot's structure, groups, and variables before campaigns send traffic to it - Keep programmatic sessions against production bots behind [human-in-the-loop](/concepts/human-in-the-loop) confirmation

Without FlowRunner

Submissions pool inside the bot Conversational forms collect rich answers that nobody exports until Friday

Broken sessions go unnoticed A webhook inside the bot fails silently and a week of respondents hit a dead end

The bot is a cul-de-sac Great qualification conversations end in the Typebot dashboard instead of the pipeline

With FlowRunner

Every submission moves within minutes Sweeps pull results on a schedule and land them in the CRM, the sheet, and the channel

Execution logs get read by an agent Result logs are checked for webhook and integration errors, and failures page a person

Conversations feed the pipeline Answers and variables map to CRM fields, and sales reads the respondent's own words

Use Case Scenarios

Qualification answers become CRM records with the transcript attached

The website's qualification typebot collects company, role, team size, and the visitor's problem in their own words. Every fifteen minutes an agent runs List Results with the stored cursor, calls Get Result for each new submission, and creates or updates the contact in [HubSpot](/integrations/hubspot) with the structured answers mapped to properties. The free-text problem description lands on the record as a note. Submissions that hit the qualification bar also post to [Slack](/integrations/slack), so the first sales touch quotes the respondent instead of guessing at them.

The silent webhook failure that pages someone before Friday

A typebot hands off to a payment link through a webhook block. One morning the endpoint starts returning errors, and respondents see a polite dead end. The nightly audit agent runs List Results for the day's submissions and calls Get Result Logs on each, scanning for failed webhook calls and integration errors. When failures cluster, it posts the affected result IDs, the failing block, and the error text to the engineering channel. The team learns about the breakage from the logs the same day, with the specific sessions that need a manual follow-up already listed, instead of discovering a quiet week of lost signups.

An agent walks the bot before every launch

Before a campaign sends traffic to a revised typebot, a verification agent exercises it end to end. It calls Get Typebot to confirm the published version contains the expected groups and variables, then Start Chat with the campaign's pre-filled variables and Continue Chat through a scripted set of replies, one call per turn, checking the bot's messages and expected inputs at each step. A transcript of the walk-through goes to the campaign owner for sign-off. The first person to discover a broken branch is an agent on Tuesday, not a prospect on launch morning.

Human-in-Loop Highlight

Start Chat and Continue Chat do not run against a sandbox. They open real sessions with a published typebot, and every block in that bot executes for real: webhook calls fire at production endpoints, integration blocks write to connected systems, and each session lands in the same results data your funnel metrics are computed from. Get Result Logs will happily show you the damage afterward. So before an agent batch-drives sessions against a production bot, whether for pre-launch verification or assisted completion, it stops and presents the plan: which typebot, which pre-filled variables, how many sessions, and which integration blocks the bot will execute. A person confirms, and cleanup of the test results gets scheduled in the same breath. The conversation is simulated. The webhooks are not.

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

7 actions

Typebots

List Typebots Lists the typebots accessible with the configured API token, optionally filtered to a single workspace, returning each typebot's ID, name and publication state. The discovery step that supplies IDs to every other operation.
Get Typebot Retrieves the full definition of a typebot by ID, including its groups, blocks, variables, settings and theme. Used to verify a bot's published structure before driving sessions or mapping answers.

Results

List Results Lists the results collected by a typebot, one per respondent session, with pagination via limit and cursor. The sweep step behind every submission-sync workflow.
Get Result Retrieves a single result by ID, including all answers and variable values captured during the session. Used to pull the full submission before syncing or escalating it.
Get Result Logs Retrieves the execution logs for a specific result, including webhook calls, integrations and errors from the respondent's session. The action that turns silent bot failures into pageable events.

Chat

Start Chat Starts a new session with a published typebot, returning the bot's initial messages and a sessionId, with optional variable pre-fill to skip questions or personalize the flow. The entry point for programmatic conversations.
Continue Chat Continues a session by sending the respondent's reply and returning the bot's next messages and expected input. Called once per turn until the typebot completes.

Frequently Asked Questions

What can FlowRunner do with Typebot?

FlowRunner agents can run List Typebots, Get Typebot, and List Results in Typebot, plus 4 more actions.

Does connecting Typebot to FlowRunner require OAuth?

No. Typebot connects to FlowRunner with an API key, no OAuth flow required.

Can Typebot trigger a FlowRunner workflow automatically?

Typebot doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.

Start building with Typebot

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

Start Building Free Talk to us