Advanced Settings Overview – Conversation AI
This article covers the advanced settings that control Conversation AI bot behavior, including the new 100-message limit, response timing, channel preferences, and safety controls.
Configure these settings to balance automation with human oversight and deliver the right customer experience at scale.
Accessing Advanced Settings
Every Conversation AI bot—whether Prompt-based or Form-based—has a dedicated Settings panel where you control message limits, response timing, and channel behavior.
Go to Sub-Account → Conversation AI → Bots.
Click the bot you want to configure, then select Bot Settings from the navigation.
Modify any field below, then click Save at the bottom of the panel.
Maximum Message Limit (1–100)
The Maximum messages a Bot can send in a Conversation field caps how many replies your bot may send in a single thread before going to sleep. HighLevel recently raised this ceiling from 25 to 100 messages, allowing bots to handle longer, more complex dialogues without manual intervention.
Previously, over 5,000 contacts hit the 25-message cap every day, causing conversations to stall. The new 100-message limit reduces manual escalations and keeps customer journeys flowing smoothly.
How It Works
Once the bot sends its maximum number of messages in a conversation, it automatically goes dormant. Only bot-generated messages count toward the limit—human agent replies, workflow messages, and inbound contact messages do not increment the counter.
Raising the limit can increase Conversation AI usage costs. Every bot-generated message counts toward your plan's AI generation quota, so monitor usage after adjusting this setting.
Setting the Limit
In the Bot Settings panel, locate the Maximum messages a Bot can send in a Conversation field. Enter any value from 1 to 100 (the default for new bots is 100). Click Save to apply the change immediately to all active conversations.
Response Delay Settings
Control how quickly your bot replies to create a more natural conversation pace. Add artificial delays to simulate human typing behavior or keep responses instant for high-volume support queues.
Set a floor (in seconds) before the bot can reply. Use 2–5 seconds to mimic human reading time.
Cap the longest delay (in seconds). Prevents frustration if the bot takes too long to respond.
Channel-Specific Controls
Enable or disable your bot on individual channels (SMS, Facebook Messenger, Instagram, WhatsApp, etc.). The Maximum Message Limit applies uniformly across all enabled channels—you cannot set different limits per channel.
If certain channels require shorter conversations, create a separate bot with a lower message limit and assign it only to those channels.
Safety and Fallback Options
Configure guardrails to protect contacts from repetitive loops or off-brand responses. These settings work alongside the message limit to ensure quality interactions.
Displayed when the bot cannot generate a confident reply or when it hits the message limit.
Block specific words or phrases to prevent the bot from discussing sensitive topics.
Automatically assign conversations to a human agent after the bot hits its limit or detects frustration.
Resetting the Message Counter
When a bot hits its maximum message limit, it goes to sleep. You can wake it manually or via workflow to resume the conversation.
Manual Reset
In the Conversations inbox, open the sleeping bot conversation and mark it as Read. This resets the counter to zero and reactivates the bot immediately.
Workflow-Based Reset
Use the Update Conversation AI Bot and Status workflow action to programmatically wake the bot. Pair this with conditions (e.g., "contact replied with keyword") to resume automation only when appropriate.
Set a workflow trigger on "Conversation Status = Bot Sleeping" to automatically re-engage the bot after 24 hours if the contact sends a new message.
Frequently Asked Questions
Was this article helpful?
That’s Great!
Thank you for your feedback
Sorry! We couldn't be helpful
Thank you for your feedback
Feedback sent
We appreciate your effort and will try to fix the article