Set up AI Email Chatbot

• From your bot's Training section, choose whether to:
  • Select an existing knowledge base, or
  • Create a new knowledge base (for example, Dave's knowledge base for a specific client or brand).

Use your preferred LLM (ChatGPT, Gemini, Claude, or similar) with a structured prompt pack or snapshot to generate knowledge base content. Include at least:

• Niche or industry.
• Business name.
• Location (full address is recommended).
• Service area or regions served.
• Business hours.
• Contact information.
• Main service or offer description.

Copy the generated content when it looks accurate and complete.

• Open the knowledge base in HighLevel.
• When asked how to add content, choose Rich text (not web crawler, FAQ, or table).
• Click Add rich text.
• Paste the content from your LLM into the editor.
• Click Save to store the knowledge base article.

• Return to the bot's Training page.
• Refresh the page if the new knowledge base does not appear immediately.
• Select your knowledge base (for example, Dave's knowledge base) from the dropdown.
• Click Save so the bot uses this knowledge when answering emails.

• In your LLM of choice, use a prompt pack to produce three distinct sections:
  • Personality: How the bot should speak (tone, formality, friendliness).
  • Goal: What the bot is trying to achieve in each conversation (for example, book appointments, qualify leads, resolve support issues).
  • Additional information: Extra guidance, such as escalation rules, what not to say, and brand nuances.

Review the output to make sure it aligns with your brand and email tone.

• In your bot's Goals or equivalent settings area, locate:
  • The Personality field.
  • The Goal field.
  • The Additional information field.
• Paste each generated section into its matching field.
• Click Save.

After saving, look for the confirmation message like bot update successfully to verify your changes applied.

Configure how the bot handles scheduling when the template or goals include booking.

• Select the relevant calendar or calendars the bot should use.
• Decide whether to check Don't book my appointment, only send booking link:
  • Leaving this unchecked allows the bot to book directly through the conversation, which is usually preferred for email flows.
• Optionally configure:
  • Pause bot after booking so the bot stops replying once an appointment is set.
  • A workflow to trigger after a booking for reminders or follow-ups.
  • Transfer to employee after booking so a person can take over if needed.
  • Whether contacts can cancel or reschedule through the bot.
  • Use multiple calendars if different offers or team members share the same email bot.

Use workflow actions when certain email conversations should kick off automation.

• Add a Trigger a workflow action.
• Select the workflow to run.
• Define when to trigger it (for example, after the contact shares certain details or reaches a specific point in the conversation).

Keep your CRM data accurate based on what contacts share over email.

• Add a Contact info or similar action.
• Map conversation data into custom fields or standard fields (for example, preferred service, budget range, subscription interests).
• Set conditions so the bot only updates data when it is confident about the values.

Prevent over-automation by defining when the bot should exit the conversation.

• Enable a Stop bot action for scenarios such as:
  • The contact asks for a human.
  • The conversation goal is fully complete.
• Configure a final message so the bot exits gracefully.
• Optionally allow reactivation if the contact sends a new email later.

Route conversations to real team members when the Email AI chatbot hits its limits.

• Turn on Human handover.
• Choose the user or internal inbox to receive the conversation.
• Define the conditions for handover, such as specific keywords, topics, or high-intent signals.

If you use multiple bots (for example, a support bot and a sales bot), transfer emails to the most relevant one.

• Configure a Transfer bot action.
• Select the target bot.
• Set the conditions that trigger the transfer, such as when a support request turns into a sales opportunity.

Auto follow-ups keep conversations moving when contacts stop replying.

• Enable Auto follow up.
• Set the delay for the first follow-up:
  • For email, choose a longer delay like 40 minutes or more so messages feel natural rather than pushy.
• Add up to 5 follow-ups spaced over several days to re-engage contacts who did not respond.

• Turn on a conversation summary preference if available.
• Use summaries for quick review, quality checks, or passing context to human agents.

When all actions look correct, click Save so the bot uses these behaviors in live email conversations.

• From the Conversational AI bot screen, use the built-in preview or test area.
• Send sample prompts and check how the bot replies as if it were in an email thread.

• Send an email from an external address to the inbox configured for your Email AI chatbot.
• In HighLevel, open Conversations.
• Use the Email filter so you only see email threads.
• Locate your test conversation and confirm:
  • The email appears in the thread.
  • The bot shows as active on the conversation.
  • The bot sends replies that match your knowledge base, goals, and actions.

If needed, you can deactivate the bot from the conversation to take over manually and adjust settings based on what you observe.