A Chatbot That Knows
When To Step Aside.

Most chatbots fail at the handover. Build branching flows with no code, call your own API mid conversation, let Meta's AI agent handle the open questions, and pass to a human the moment intent changes.

No code builderAPI call mid flowAI fallbackClean human handover
01

Deflect the repeats

Keyword, intent and menu driven answers handle the questions you get all day.

02

Real data in the answer

A mid flow API call means the bot can quote a real order status, not a canned line.

03

No dead ends

When the bot cannot help, a human takes over with the whole thread visible.

The builder

Drag the branch, test it, publish it

Triggers cover keywords, intents, button taps, campaign replies and inbound events. Steps cover messages, questions, Flows, conditions, delays, tag writes and HTTP calls to your systems.

Every flow can be tested in a sandbox against a real number before it goes live, and the run log shows exactly which branch a conversation took.

  • Keyword, intent and event triggers
  • Branching conditions on contact fields
  • HTTP step to call your own API
  • Sandbox testing and per run logs
Flow · Lead qualification1,284 runs today
StepDetailState
TriggerInbound messagelive
AskBudget quick replylive
Branch1 Cr and abovelive
HandoverSenior repon intent
Deflected 62% · median run 14s4 branches

How it works

From zero to live,
in four steps.

01Pick a triggerA keyword, an intent, a button tap or an inbound event.
02Build the branchesQuestions, conditions and Flows, with no code.
03Add your dataAn HTTP step that reads or writes in your systems.
04Set the handoverConditions that pass the chat to a person.

What is inside,
in detail.

Trigger types

Keywords, intents, buttons, campaign replies and API events.

Conditions

Branch on contact fields, tags, time or previous answers.

HTTP step

Call your API mid conversation and use the response.

AI fallback

Meta's agent answers what the flow did not anticipate.

Handover rules

Intent, sentiment or explicit request routes to a human.

Run logs

Every execution, step by step, for debugging.

Where rivals say no

The same feature,
across four platforms.

FeatureWBIZWATIInteraktAiSensy
HTTP call inside a flowHigher tierLimited
Sandbox testing before publish
AI fallback with handoverPartialPartial
Per run execution logLimited
Included on the standard planPartialPartialPartial

Competitor entries are indicative, from public 2026 pages. See the full 89-row benchmark.

FAQ

WhatsApp chatbot questions

Anything missing, a founder will answer directly.

No. The builder is drag and drop. The HTTP step is there if you want your own data, and a developer can help with just that piece.

Yes, through Meta's AI agent as a fallback, configured with your FAQs, files and website.

On intent, sentiment or an explicit request, the conversation is assigned to a human with the full thread visible.

Yes, in a sandbox against a real number, with run logs for each execution.

Yes, automations are part of the plan rather than an add on.

Explore more

The rest of the platform

WhatsApp Flows

Structured forms the bot can open mid conversation.

See Flows

Build a chatbot that hands over properly

Start free, publish one qualification flow, and watch how much it deflects.