Setting Up Conversation AI in Captivation Hub — A Step-by-Step Guide

This guide walks you through configuring Conversation AI inside your Captivation Hub sub-account from scratch — what the feature does, how to spin up your first bot, and the settings worth tuning before you put it in front of customers.

What Conversation AI Actually Does

Think of Conversation AI as your always-on communications layer. It's an automated messaging system that uses AI bots to answer questions, deliver support, and keep conversations moving across the channels your contacts already prefer — without making your team copy and paste the same answers all day.

Why Teams Lean On It

• Multi-bot architecture — Run several bots side by side, each tuned for a different conversation type (general support, booking, lead qualification, and so on).
• One designated Primary bot — Pick a single bot to catch every inbound message that isn't already being handled by a workflow. Keeps coverage simple.
• Workflow-aware — Drop bots inside automations to deliver context-specific replies at exactly the right step.
• Channel coverage — SMS, Email, Facebook, Instagram, and Live Chat are all on the table.
• Test as you go — A built-in chat window lets you trial the bot, leave feedback, and iterate without affecting any live contacts.
• Smarter appointment flows — Bots can send booking links, hand off to another bot, or kick off a workflow once an appointment lands.
• Improved transcript handling — Better parsing of prior messages is on by default. No toggles to flip; just better answers.
• Custom object names — Renamed Contacts to Clients? The bot UI follows your naming, so your team isn't translating in their head.

FYI: The transcript-handling improvements apply automatically to every active bot — there's nothing to switch on.

Step 1 — Build Your First Conversation AI Bot

Spinning up a new bot inside Captivation Hub is fast. Here's the path:

1. Click "Create Bot" in the Conversation AI dashboard.
2. Pick a Prompt Template. Templates are a head-start, especially if AI prompting is new to you. Everything inside the template is fully editable after creation:
   • General Q&A Template — Pre-shaped for support and general inquiries.
   • Appointment Booking Template — Pre-shaped for scheduling conversations.
3. Set the Bot Status. This decides whether — and how — the bot replies to live messages:
   • Off — Inactive in conversations. Useful while you're still building.
   • Suggestive — Drafts replies inside the composer for your team to send or edit.
   • Auto-Pilot — Replies on its own, no manual step required.
4. Assign Channels. In the bot settings, toggle on the channels (SMS, Email, Facebook, Instagram, Live Chat) you want this bot covering.
5. Designate as Primary (optional). Only one bot per sub-account can be Primary. The Primary bot fields any incoming conversation that isn't already routed by a workflow.

Worth knowing: Choosing a Bot Template only seeds the Prompts for your bot. You can rewrite or replace any of it after the fact — templates aren't locked.

Step 2 — Decide Between Primary and Non-Primary Bots

Both bot types matter, but they play different roles. Setting up channels correctly on each is what makes the whole system run cleanly.

Primary Bot

• Handles every general inbound conversation that hasn't been picked up by a workflow.
• Should have all of your "main" channels assigned so nothing slips through.

Non-Primary Bots

• Pick channels based on the workflows you'll point them at.
• Make sure the workflow's channels match what's enabled on the bot — otherwise the handoff fails silently.

Step 3 — Tune the Advanced Settings

Advanced Settings are where you fine-tune timing, message limits, and behavior so the bot feels natural in real conversations:

• Wait Time Before Responding — Add a small delay so replies don't fire instantly (which often reads as robotic).
• Maximum Message Limit — Cap the number of messages the bot will send in a single thread before stopping or handing off.
• Send Bot to Sleep — Temporarily silence the bot during manual takeovers or workflow-triggered actions so it doesn't talk over itself.

Tip: If you've renamed standard CRM objects, head to Settings → Objects at the sub-account level. Conversation AI will display whatever names you've set.

Step 4 — Train the Bot

Training is what separates a generic bot from one that actually sounds like your business. Two tools handle most of the work:

• Web Crawler — Point it at URLs, full domains, directories, or even Google Docs. The crawler ingests the content so the bot can pull from it when answering.
• Custom Bot Responses (FAQs) — Lock in exact phrasing for the questions you want answered consistently. When a contact asks something close to a stored FAQ, the bot uses your wording verbatim.

Step 5 — Set the Bot's Goals

Goals shape personality, intent, and any extra context the bot needs to keep on-brand. Think of this as your bot's job description.

Prompt

The Prompt is where personality, intent, and additional information come together. It's the single biggest lever for whether the bot feels right to your audience.

Form-Based (Guided Form) note: The Additional Instructions field is the place to lean on for guided-form bots — that's where you specify how the bot should behave between form steps.

Personality defines the bot's tone. A few examples:

• Friendly — Casual, approachable, conversational.
• Professional — Polished and business-forward.
• Formal — Reserved and tightly structured.

Intent is the primary objective — what the bot is trying to accomplish:

• Resolving Queries — Focus on answering the contact's question.
• Generating Leads — Steer conversations toward conversion or capture.

Additional Information is where you add anything else — specifics about your business, edge cases the bot should handle differently, products the bot should mention, etc.

Appointment Booking Actions

If the bot is booking calls or meetings, you've got several actions to choose from:

• Book Appointment — Selects a calendar from the ones in your sub-account.
• Send Booking Link — Skips the in-thread time-slot picker and just delivers a calendar link. Useful when you want contacts to choose on their own.
• Pause After Booking — Mutes the bot once an appointment lands, keeping things tidy and allowing for a clean handoff.
• Transfer to Employee — Hands the conversation off to another AI bot. Works well when a general support bot needs to pass a contact to a booking-specific bot, or vice versa. The handoff is seamless — the new bot picks up the thread without any gap.

Step 6 — Test and Refine in Real Time

Before any live conversations happen, take advantage of the testing tools:

• Chat With the Bot — Free, unlimited test conversations let you stress-test how the bot handles different questions.
• Leave Feedback on Replies — A thumbs-up reinforces good behavior. A thumbs-down automatically generates a new FAQ entry in the training section, so you can correct the bot in place.
• Edit the Prompt Live — The pencil icon next to the feedback buttons opens a quick edit on personality, intent, or additional information without leaving the test screen.
• Reset the Conversation — Start fresh with one click — no browser refresh needed — so you can validate edits immediately.

Where to Go From Here

A solid post-setup checklist:

• Confirm which bot is set as Primary and that it covers every channel you want auto-handled.
• Build out additional bots for any workflow-specific scenarios (booking, qualification, etc.).
• Spend a few days inside Bot Trial to find the rough edges before flipping the bot to Auto-Pilot.

Frequently Asked Questions

Do I need to flip a setting to enable the new transcript handling?

No — it's already on for every active bot in your sub-account. Nothing to configure.