FlowRunner
PricingContact
Theme
Start Free

ChatBot

AI

Drive ChatBot.com conversations and bot operations from your flows. Agents send messages to published stories, train bots on unmatched phrases and custom text, manage entities, segments, and webhooks, and pull chat archives and analytics reports.

37 actions API key available
ChatBot website ↗ Platform Documentation ↗ Capability data verified 2026-07-27
Weekly bot-improvement review kicks off on a schedule
Agent pulls the unmatched-phrase queue with List Training Phrases
Agent clusters the phrases and matches each cluster to an existing User Input interaction
Agent drafts a training plan that separates real knowledge gaps from noise
Plan posted to the bot owner in Slack with phrase counts per cluster
Owner approves the mapping; agent runs Train Phrases on approved clusters and Ignore Phrases on the rest

What This Integration Enables

A ChatBot.com story is only as good as what happens to the questions it cannot answer. This integration gives FlowRunner agents the full surface around the bot: driving conversations against a published story with Send Message, managing the story lifecycle, maintaining the entities and segments that make answers personal, and pulling chat archives and analytics into the systems where decisions get made. The unmatched-phrase queue, the part most teams never look at, becomes a managed improvement loop instead of a graveyard. The stance behind the design is simple: the bot owns the front line, and a person owns what the bot is allowed to become. Agents prepare training plans, sync visitor profiles, and route unresolved conversations onward, but changes to a production bot's behavior run through [human-in-the-loop](/concepts/human-in-the-loop/) approval. Autonomous on the routine, accountable on the rest.

Without FlowRunner

Unmatched questions pile up Phrases the bot missed sit unreviewed in the training queue for weeks
Training by gut feel Whoever opens the dashboard decides what the live bot learns, with no review step
Escalations lose the transcript Conversations the bot cannot resolve reach a person stripped of context

With FlowRunner

A standing training loop Every unmatched phrase gets clustered, mapped, and either trained or dismissed
Approved changes only The production bot learns nothing until the owner signs off on the mapping
Escalations arrive with history The full transcript and visitor profile travel with every handoff

Use Case Scenarios

The weekly training loop

Every week, an agent pulls the unmatched phrases with List Training Phrases, clusters them by intent, and matches each cluster to an existing User Input interaction in the story. It drafts a plan: these forty phrases map to the returns interaction, these twelve need a new answer, these nine are noise. The bot owner reviews the plan in [Slack](/integrations/slack), approves the mappings, and the agent runs Train Phrases on the approved clusters, Train Custom Text where a new FAQ answer is needed, and Ignore Phrases on the noise. The bot gets measurably smarter every week, and nobody trained it by accident.

Escalation that keeps the conversation

An agent drives a support conversation through Send Message with a stable session identifier. When the visitor's question falls outside what the story can resolve, the agent pulls the full transcript with Get Chat and the visitor's profile with Get User, then opens a ticket in [Zendesk](/integrations/zendesk) carrying both. The support rep picks up mid-conversation instead of starting with "how can I help you today" for the second time. The visitor never repeats themselves.

Qualified leads and the weekly scorecard

When a visitor qualifies during a bot conversation, the agent reads their attributes with Get User, syncs them into [Intercom](/integrations/intercom) as a contact for follow-up, and adds them to the right segment with Add Users To Segments so future conversations start personalized. On a schedule, the agent calls Get Report for conversation and engagement metrics and appends the numbers to a scorecard in [Google Sheets](/integrations/google-sheets), so the team judges the bot on data rather than anecdotes.

Human-in-Loop Highlight

Delete Story permanently removes a bot and every interaction built into it, and it cannot be undone. So when a bot consolidation flow identifies a story that looks retired, the agent does not clean it up on its own. It inventories the account with List Stories, checks the story's recent traffic with Get Report, and posts to the bot owner in Slack: "Story 'Onboarding v1' has had no conversations this quarter and 'Onboarding v2' is published. Delete it? This removes the story and all of its interactions permanently." Only an explicit yes runs Delete Story. Months of interaction design should never disappear because an agent guessed which bot was the old one.

Agent processes routinely
Detects exception requiring judgment
Clear match Continues automatically
Ambiguous Routes to human via Slack
Human decides
Agent resumes with decision

Agent Capabilities

37 actions

Conversations

1
  • Send Message Sends a visitor message to a published story through the Chat API and returns the bot's response interactions. A stable session identifier keeps multiple turns in the same conversation context, so agents can drive full conversations programmatically.

Stories

5
  • List Stories Lists all stories on the account with ID, name, version, published state, and enabled integrations. The discovery step before any story-scoped operation.
  • Get Story Retrieves a single story by ID, including its published state and integrations.
  • Create Story Creates a new empty story ready for training and interactions.
  • Update Story Updates a story's name or description.
  • Delete Story Permanently deletes a story and all of its interactions. Cannot be undone. Routed through human approval in FlowRunner workflows.

Training

5
  • List Training Phrases Lists the phrases visitors typed that the bot could not confidently match, with how often each one missed. The raw material of the improvement loop.
  • Train Phrases Trains unmatched phrases to a target User Input interaction, teaching the bot to match those phrases to that response.
  • Train Custom Text Trains a story on free-form text such as an FAQ or knowledge base entry, grounding the bot in written knowledge quickly.
  • Ignore Phrases Dismisses irrelevant unmatched phrases so training suggestions stop surfacing them.
  • Delete Training Phrases Removes unmatched phrases from the training queue entirely, keeping training data clean.

Entities

5
  • List User Entities Lists the custom entities the bot recognizes in conversations, such as product names or cities, each with values and synonyms.
  • Get User Entity Retrieves a single entity with all of its entries and synonyms.
  • Create User Entity Creates a new entity so the bot recognizes its values and synonyms in conversations.
  • Update User Entity Updates an entity's name or replaces its entries.
  • Delete User Entity Permanently deletes an entity and removes it from interactions that referenced it. Cannot be undone.

Users

8
  • List Users Lists chat visitors with their attributes, conversation counts, and ban status.
  • Get User Retrieves a single visitor with full attributes, conversation history, and segment memberships. Used when qualifying leads or escalating with context.
  • Create User Pre-populates a visitor profile with attributes your bots can reference.
  • Update User Merges new attributes onto an existing visitor profile.
  • Delete User Permanently deletes a visitor and their profile data. Cannot be undone.
  • Update User Segments Adds, replaces, or removes a single user's segment memberships.
  • Set User Ban Bans or unbans a visitor from interacting with your bots.
  • Export Users Exports visitors and selected fields as CSV for offline analysis or backup.

Segments

5
  • List Segments Lists user segments with member counts. Segments group visitors for targeting and reporting.
  • Create Segment Creates a new segment for targeting campaigns or personalizing flows.
  • Add Users To Segments Places specific users into segments regardless of automatic conditions.
  • Update Segment Renames a segment.
  • Delete Segment Permanently deletes a segment grouping. Users themselves are untouched. Cannot be undone.

Webhook Connections

5
  • List Webhook Connections Lists the webhook connections that tie stories to external callback URLs.
  • Get Webhook Connection Retrieves a single connection with its target URL, headers, and error state.
  • Create Webhook Connection Wires a story's webhook blocks to an external URL, with optional token, headers, and basic auth.
  • Update Webhook Connection Updates a connection's URL, token, headers, or auth.
  • Delete Webhook Connection Permanently removes a connection so it stops receiving callbacks. Cannot be undone.

Archives and Reporting

3
  • List Chats Lists archived conversations with transcripts, sources, and timestamps, paginated by cursor.
  • Get Chat Retrieves one archived conversation with the full visitor-and-bot transcript. The context payload for escalations.
  • Get Report Pulls analytics for a chosen metric over a date range: total conversations, messages, busiest periods, interaction popularity. Feeds dashboards and the training loop's before-and-after measurement.

Frequently Asked Questions

What can FlowRunner do with ChatBot?

FlowRunner agents can run Send Message, List Stories, and Get Story in ChatBot, plus 34 more actions.

Does connecting ChatBot to FlowRunner require OAuth?

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

Can ChatBot trigger a FlowRunner workflow automatically?

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

Start building with ChatBot

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