Conversation AI in HighLevel

Most businesses lose leads simply by being slow to respond. Conversation AI removes that problem entirely.

The bot is always on, always available, and trained on your business information from day one.

Reading time: approximately 10 minutes.

What Is Conversation AI?

Conversation AI is HighLevel’s built-in AI bot for handling inbound text-based messages. It lives inside the sub-account settings and connects directly to the Conversations inbox.

When a contact sends a message on a supported channel, Conversation AI reads the message, understands the intent, and either suggests a response for a human to review or sends one automatically depending on the mode you have configured.

It is part of the AI Employee suite, which also includes Voice AI, Reviews AI, Funnel AI, and Content AI. Each AI Employee tool handles a different type of business interaction – Conversation AI specifically handles the written back-and-forth with contacts across messaging channels.

How It Works

The bot is trained on information you provide. That training comes from two sources: the Web Crawler, which pulls content from your website URLs or Google Docs, and Custom Bot Responses, which are specific question-and-answer pairs you define manually.

When a contact sends a message, the bot uses the last 10 conversations or up to 800 words of context to generate a response that fits the conversation. It does not just match keywords – it understands the flow of the conversation and adjusts its tone and phrasing based on what has already been said.

The bot has two primary intents you configure at setup:

• General Support: The bot answers questions and handles inquiries based on its training data. Good for customer service, FAQs, and qualification conversations.
• Appointment Booking: The bot asks a series of configurable qualifying questions and then books an appointment directly on a calendar you specify. If the contact already has an active or confirmed appointment, the bot automatically switches to General Support mode until the appointment is completed or cancelled.

Both intents can run in either Suggestive or Auto-Pilot mode. Training and testing in the Bot Trial tab are completely free before deploying to any live channel.

Suggestive Mode vs Auto-Pilot Mode

Suggestive Mode

In Suggestive mode, the bot drafts up to three response options and displays them inside the message composer box when a user opens that conversation in the Conversations inbox.

The user reviews the suggestions and can send one as-is, edit it before sending, or ignore all three and write their own reply. The AI assists but the human controls every message that actually goes out.

This is the right starting point if you are new to the bot. It lets you monitor how the AI is responding before you hand it full control, and it gives you a chance to catch any training gaps early.

Auto-Pilot Mode

In Auto-Pilot mode, the bot responds to inbound messages automatically without any human involvement. When a message arrives, the system waits two minutes before sending a response.

This delay accounts for contacts who send multiple messages in quick succession – the bot reads all of them together before responding rather than replying to each one individually.

The bot varies its question phrasing to avoid sounding repetitive, adjusts tone based on how the contact has been communicating, and gently redirects off-topic replies back to the main intent of the conversation.

Auto-Pilot is designed for after-hours coverage, high-volume inbound, or any situation where a human cannot realistically respond to every message in real time.

What You Can Build With It

• 24/7 lead response that engages every inbound inquiry immediately regardless of the time of day, eliminating the response delay that causes most lead drop-off
• Automated appointment booking that takes a contact from first message to confirmed calendar slot without any human involvement, including sending a direct booking link or displaying available calendar slots
• Multi-channel coverage with a single bot trained once and deployed across SMS, web chat, Facebook Messenger, Instagram DMs, and Google Business Profile simultaneously
• Human-assisted responses in Suggestive mode that speed up the response process for your team by pre-drafting replies they can review and send in seconds
• Specialized multi-bot setups where different bots handle different services or departments, each trained on its own knowledge base, with bot transfer nodes routing contacts to the right bot mid-conversation
• Image-aware conversations where the bot reads and responds to images shared by contacts, such as screenshots, uploaded documents, or photos relevant to the inquiry
• Workflow integration through the Flow Builder, connecting the bot conversation directly to HighLevel workflows, webhooks, and automated follow-up sequences triggered by what happens in the chat
• Clean human handoffs using the Maximum Message Limit, Send Bot to Sleep, and Transfer to Employee settings to transition contacts to a human agent at the right moment without the bot interfering
• White-label AI for agency clients with custom bot names, branding, and rebillable usage costs that create a recurring AI service revenue line for agencies on the $497 plan

Key Definitions

Use Cases by Industry

Marketing Agencies

An agency deploys Conversation AI on behalf of a client running Facebook lead ads. Every lead that comes in through Facebook Messenger receives an immediate AI response that qualifies them with two or three questions and then books a consultation call.

The agency sets the bot up once, trains it on the client’s service offering, and connects it to the client’s appointment calendar. From that point, the bot handles every inbound Facebook message automatically without the client needing to monitor their inbox in real time.

Outcome: Every Facebook lead gets a response in under two minutes at any hour, and the calendar fills without the client touching the conversation.

Home Services Businesses

A plumbing company connects Conversation AI to their Google Business Profile and SMS. When a homeowner messages through Google asking about a service call, the bot responds immediately, asks a few questions about the issue, and books an appointment on the technician’s calendar.

After hours, the bot handles everything that would previously have gone to voicemail. No calls missed.

No leads lost because nobody was available to respond at 9pm.

Outcome: The business captures after-hours inquiries that would previously have been lost, converting them to booked jobs without any staff involvement.

Real Estate Agents

A real estate agent installs the chat widget on their website and connects Conversation AI to it. The bot is trained on the agent’s current listings, service areas, and buyer process.

When a visitor starts a chat asking about a listing or the buying process, the bot handles the conversation, answers questions from the training data, and eventually moves qualified visitors toward booking a showing or a buyer consultation call.

Outcome: Website visitors who would have bounced without getting an answer are now converted into booked appointments, even when the agent is unavailable.

Med Spas and Wellness Clinics

A med spa deploys Conversation AI on SMS and Instagram. The bot handles inbound inquiries about treatments, pricing, and availability.

It qualifies the contact and books consultations directly on the provider’s calendar.

A specialized bot is trained specifically on each treatment type, with Transfer to Bot routing contacts to the right specialist bot based on what they are asking about. The main bot handles the initial message and routes accordingly.

Outcome: Each contact gets accurate, service-specific information without the front desk handling every inquiry manually, and bookings happen around the clock.

Coaches and Course Creators

A business coach uses Conversation AI in Suggestive mode on their website chat. When a potential client messages asking about coaching packages, the bot drafts a personalized response for the coach to review and send.

The coach stays in control of every message while saving the time it would take to write responses from scratch. The AI handles the draft.

The coach adds their personal touch and hits send. Response time drops from hours to minutes.

Outcome: The coach’s response time improves dramatically without losing the personal touch that matters in high-ticket sales conversations.

Who Is This For?

How to Set Up Conversation AI

This walkthrough covers enabling the feature at the agency level, creating and training a bot, selecting channels and mode, testing before going live, and monitoring early performance.

Step 1: Enable AI Employee at the Agency Level

Switch to Agency View using the switcher at the top left of the sidebar. Go to Settings, then Company.

Scroll down to the AI Employee section and toggle it on.

This is the master switch. Without it, Conversation AI settings will not appear inside any sub-account.

Step 2: Enable Conversation AI in the Sub-Account

Switch into the sub-account where you want to deploy the bot. Go to Settings, then Conversation AI.

Toggle on the feature for this location.

If you are an agency and want to rebill usage to this client, go back to your Agency account, find the sub-account under the Accounts tab, click the three dots, select Manage Client, go to the Rebilling tab, find Conversation AI, toggle it on, and set your markup with the slider.

Step 3: Create Your Bot

Inside the sub-account Conversation AI settings, click Create Bot or go to Agents List and click Create Bot. Give the bot a descriptive name.

Select the bot’s intent: General Support for answering questions, or Appointment Booking for guiding contacts through a booking flow. If you choose Appointment Booking, select the calendar to use from the dropdown and configure whether to show available slots or send a booking link.

Step 4: Train the Bot

Go to the Bot Training tab. Under Web Crawler, enter your website URL or specific pages.

The crawler will index that content and add it to the bot’s knowledge base. You can also add Google Docs URLs for internal documentation.

Under Custom Bot Responses, add FAQ pairs for your most common questions. Write each question the way a customer would actually ask it.

Write the answer exactly as you want the bot to deliver it. Add as many pairs as your business needs.

Both training methods are free. Take time here – the quality of the training determines the quality of the responses.

Step 5: Configure Advanced Settings

In the Advanced Settings section, configure:

• Wait Time Before Responding: The default two-minute delay in Auto-Pilot mode. Adjust if needed for your use case.
• Maximum Message Limit: How many bot messages are sent before the conversation is handed off or the bot stops.
• Pause After Booking: Enables a clean stop after an appointment is successfully booked.
• Send Bot to Sleep: Pauses the bot automatically when a human team member sends a manual reply in the conversation.

Step 6: Select Channels

In the Supported Channels dropdown, select which channels this bot is active on: SMS, web chat, Facebook Messenger, Instagram DMs, Google Business Profile, or any combination.

You can run the bot on all channels at once or start with one channel while you refine the training.

Step 7: Choose the Mode

Select Suggestive if you want AI-drafted replies that a human reviews and sends. Select Auto-Pilot for fully automated responses without human involvement.

If this is your first deployment, start with Suggestive. Run it for a week, review the responses, improve the training, and then switch to Auto-Pilot once you are confident the bot is performing well.

Step 8: Test in Bot Trial

Before going live on any channel, go to the Bot Trial tab. Select the intent from the dropdown and type messages as if you were a customer.

Test your most common inquiries, edge cases, and the booking flow.

Identify any gaps in responses and go back to Bot Training to add more FAQ pairs or additional Web Crawler URLs. Bot Trial is completely free and has no limits on how many times you test.

Step 9: Deploy and Monitor

Save the bot configuration. Send a live test message from a real device on one of your enabled channels to confirm the bot responds correctly.

Monitor early conversations in the Conversations inbox. Click into any AI-generated message and use Response Info to see which prior messages the bot used as context for that reply.

Use this to diagnose any unexpected responses and improve the training accordingly.

How It Connects to the Rest of HighLevel

• Conversations Inbox – all Conversation AI interactions appear in the unified Conversations inbox alongside messages from every other channel. Suggestive mode responses show directly in the message composer. Human team members can take over any AI-managed conversation at any time.
• Calendars and Appointments – the Appointment Booking intent connects directly to any calendar in the sub-account. The bot checks availability and books confirmed appointments without the contact leaving the conversation. Multi-calendar support means different bots or flow paths can route to different team calendars.
• Workflow Builder – Conversation AI connects to workflows through the Flow Builder’s workflow trigger nodes. Bot conversations can trigger GHL workflows at specific points, passing the contact into nurture sequences, tag-based automations, or pipeline updates based on what happened in the conversation.
• Contact Management – the bot interacts with contacts already in the CRM and creates new contact records when an unknown number or user messages in. Contact data captured during the conversation, such as name, phone, or service interest, can be written back to the contact record.
• Voice AI – Conversation AI handles text channels while Voice AI handles phone calls. Together they create a complete inbound coverage system. A contact might engage through SMS with Conversation AI and then receive a follow-up call from Voice AI as part of the same workflow sequence.
• Chat Widget – adding Web Chat as a supported channel in Conversation AI automatically activates the AI inside the HighLevel Chat Widget installed on your funnel or website. No separate configuration is needed for the widget itself beyond adding the channel in the bot settings.

Common Questions

To Wrap It Up

Conversation AI solves one of the most consistent problems in small business sales: slow response time. Most leads go cold not because the business is bad at what it does, but because nobody was available to respond quickly enough.

The bot removes that constraint entirely.

The feature is more capable than most people expect the first time they look at it. Between the Flow Builder’s visual canvas, multi-bot routing, image interpretation, multi-calendar booking, and deep integration with the rest of HighLevel, it is a genuine automation layer for customer communication – not just a basic chatbot.

The most important thing to get right is the training. A well-trained bot that knows your business, your services, and your most common questions will perform dramatically better than a bot deployed in five minutes with no training investment.

1. Enable AI Employee at the agency level before looking for Conversation AI settings inside a sub-account
2. Start with Suggestive mode, review the AI’s responses for a week, and switch to Auto-Pilot once the training is solid
3. Use the Web Crawler to pull in your website content and add FAQ pairs for your most important questions before testing anything
4. Run every test scenario in Bot Trial before activating the bot on a live channel
5. Check Response Info on early conversations to see what context the bot used, and use that to catch and fix training gaps quickly

A well-trained Conversation AI bot running in Auto-Pilot across your key channels means no lead goes unanswered and no appointment opportunity is missed – regardless of what time it comes in.