Skip to content
ManySetter
Guide · Manychat AI

Manychat AI: what it does and when to add a sales setter

Manychat already has real AI features. Here’s what each one does, where it works and how to tell if it’s enough for your sales.

ManySetter teamUpdated 9 min read

On this page
  1. 01The features
  2. 02Where each works
  3. 03Set it up
  4. 04What it does well
  5. 05Native AI vs a setter
  6. 06One test, both tools
  7. 07Messaging rules

Chapter 01The features

Which Manychat AI feature should you use?

Manychat AI isn’t one switch. Each feature does a different job, with its own setup:

  • AI Replies and Knowledge

    Answer Instagram questions from your Knowledge library, in your voice. Can’t answer? Assign a person.

  • AI Goals

    After a reply: share a link, encourage a follow or capture lead details into fields.

  • AI Step

    An AI part inside a flow, with a goal, context, tasks and a reply saved to a field.

  • Intention Recognition

    Routes a message by meaning, not an exact keyword, into the right workflow.

Two creation tools help you build, not reply: the Flow Builder Assistant drafts automations and Text Improver rewrites messages.

Paid, and partly in beta

AI Replies, Knowledge and Goals belong to Manychat’s paid Instagram AI, which Manychat describes as beta. Check what your account shows.

Chapter 02Channels

Where each Manychat AI feature works

Manychat AI by channel

FeatureInstagramMessengerWhatsApp
AI Replies and KnowledgeBest for: Repeated Instagram questionsAvailableNot documentedNot documented
AI GoalsBest for: An objective after the replyAvailableNot documentedNot documented
AI StepBest for: A defined AI step in a flowAvailableAvailableAvailable
Intention RecognitionBest for: Routing a request by meaningAvailableAvailableAvailable
Per Manychat’s documentation, September 2026. AI Step also runs on TikTok and Telegram, Intention Recognition on Telegram. A dash means not documented.

Check your own account

Availability depends on your plan, region and beta access. Confirm a feature appears in your account before you plan a funnel around it. Triggers by channel: Manychat automation guide.

Official sources: Manychat AI overview (opens in a new tab), AI Step (opens in a new tab) and Intention Recognition (opens in a new tab).

Chapter 03Set up

How to set up Manychat AI in five steps

  1. Pick the conversation outcome

    Write one outcome first: “Understand the need, answer the offer question, see if a call would help.” List only what that decision needs. Ask for everything at once and it feels like a form.

  2. Give AI accurate context

    Where to click: Manychat AIKnowledge

    Add short facts: the offer, who it helps, the pricing rules you use, objections and how to get support. Link a public page only if it holds the answer. Using AI Step? Set its goal and boundaries in the Flow Builder.

  3. Connect the next action

    On Instagram, turn on AI Replies, then pick an AI Goal if sharing a link or capturing details fits. In a flow, choose the field that stores the reply and who gets the result. Offer the booking link on fit, not by default.

  4. Test the edge cases

    Send the five messages below, then try other phrasing, a returning contact and two quick messages in a row. Check the saved fields, the real inbox and the next action, not just the tone.

0 of 5 completed

  1. Publish and review outcomes

    Where to click: PreviewSet Live

    Connect the trigger to AI Step or turn on the Instagram AI scenarios, preview, then go live. From real transcripts, track useful replies, qualified leads, calls proposed, calls booked and human takeovers.

Official sources: AI Replies (opens in a new tab), AI Knowledge (opens in a new tab) and AI Goals (opens in a new tab).

Chapter 04Native AI

What Manychat AI already does well

Manychat is a strong start when your team already uses its triggers and flows (new to them? Read the Manychat chatbot guide). Its native AI can:

  • Answer questions from your Knowledge
  • Capture contact details into fields
  • Route intents to the right workflow
  • Save replies inside a flow
  • Assign conversations to people

Choose on outcomes, not feature lists

Manychat’s AI keeps evolving. Don’t pick a second tool from an outdated list of missing features: pick it for the work that still takes time or the outcome you can’t measure.

Chapter 05Compare

Manychat AI vs a setting model

Two jobs, not a winner. Manychat AI answers and routes inside your flows. A setting model runs the sales conversation after the first reply.

Manychat AI and ManySetter’s setting model

Criterion
Data it usesManychat AI: Your Knowledge library; for AI Step, a goal and business contextManySetter: Your sales page, offer doc, FAQ, posts and past DMs, plus the lead’s source
InstructionsManychat AI: AI Step: goal, context and tasks. AI Goals: share a link or capture detailsManySetter: Offer, fit rules, voice and handoff rules. Change any reply in one sentence
TestingManychat AI: Preview, then Set LiveManySetter: Free, unlimited test conversations, never sent to a lead
Human controlManychat AI: Can assign a person when it can’t answerManySetter: Hands off when asked or unsure. The AI pauses; your team gets the context
Memory and lead stateManychat AI: Replies and details saved to fields you pickManySetter: A state per lead: criteria met, not met or unknown; qualified or booked
LearningManychat AI: Not documentedManySetter: Learns the rules you teach from tests and live chats, inside this setter only
Follow-upsManychat AI: Not documented for the AI featuresManySetter: Text or voice notes in your voice, 1 to 23 hours later. They stop on reply or booking
BookingManychat AI: An AI Goal can share a link, like your calendarManySetter: Books the call itself in Cal.com, iClosed or paid Calendly, or sends your link. Booked only when your calendar confirms it
Manychat column: Manychat’s Help Center, checked September 28, 2026. “Not documented” means we found no documentation, not that it can’t. ManySetter column: our product.

Native AI already covers your job? Keep it and skip a second subscription.

Chapter 06Test

Run one identical test

Send the same message to both setups from a test account. Write the reply you’d want before you send it.

The test message
  1. Lead

    Saw your Reel. Is the program worth it with 3 clients? How much?

  2. A good reply

    The price from your offer, a straight fit answer, then one question

What each needs to answer it well

For a good reply
The priceManychat AI: Written in Knowledge, or in the AI Step contextManySetter: In your offer: $1,500 or 3 × $550
The fit answerManychat AI: Who it’s for, in Knowledge or the AI Step goalManySetter: A fit rule: coaches with 2 to 10 clients
One next questionManychat AI: An AI Step task, or an AI Goal that captures itManySetter: Asked when the fit rule needs it
The callManychat AI: An AI Goal that shares your calendar linkManySetter: Offered once the fit rule is met, then booked in your calendar
AI Replies, Knowledge and AI Goals are documented for Instagram. On other channels, use AI Step.

Then keep the conversation going and check both:

0 of 6 completed

Chapter 07Rules

Messaging rules still apply

AI doesn’t override channel policy. Instagram, Messenger and WhatsApp keep their messaging windows, and a comment-triggered private message has its own opt-in behavior. Choosing an AI for WhatsApp sales? Run the six tests for a WhatsApp AI chatbot.

  1. 01Instagram messageArrives in Manychat
  2. 02Your automationsChecked first
  3. 03AI scenarioIf none applies
  4. 04Your teamWhen a person is needed
For some Instagram AI scenarios, Manychat first checks whether another automation can handle the message.

One owner per conversation

Using a Default Reply or several bots? Test who owns the conversation so two systems don’t answer at once.

Build triggers and follow-ups around the person’s real interaction: a follower or commenter isn’t a subscriber you can message anytime (see how the opt-in works in the Manychat Instagram automation guide). Official source: messaging windows (opens in a new tab).

ManySetter

A setting model built for sales calls.

Compare it with Manychat’s native AI: it learns your offer and voice, follows up and books the call.

  • Learns your offer and voice
  • Follows up by text or voice
  • Books calls in your calendar

Build and test free · Manychat billed separately

FAQ

Manychat AI questions

Add a setter to your Manychat conversations

Keep Manychat’s AI for what it covers. Add a setting model that learns your offer, follows up and books calls.

Use native AI first for the jobs it already covers.