> ## Documentation Index
> Fetch the complete documentation index at: https://docs.assistable.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Diagnose slow or missing GHL chat replies

> Separate the configured wait from routing, sleep, tool execution, and message delivery.

For GHL chat, **Wait time** controls the initial delay before the assistant starts a reply. Additional messages within that pending window do not restart the timer. The assistant reads the conversation history when its reply begins, so those messages can be included in the same response.

The configured wait is not a promise that a message will be delivered at that exact time. Queueing, model generation, tools, and the messaging provider add time.

## Check the setting

1. Open the intended assistant's **Chat Settings**.
2. Review **Wait time**, measured in seconds. Choose a value appropriate to how customers send messages.
3. Wait for **Saved**. If an error appears, reopen the settings and verify the stored value before testing.
4. Send one test message and record the send and receipt times. Repeat with a short burst of messages and compare the result.
5. Test a simple question and a question that calls a tool separately. This helps isolate a slow external action.

Use the [Chat Settings reference](/v3/guides/chat/activate-ghl-chat#chat-settings-reference) for every wait, sleep and channel option, including their allowed values. Diagnose the saved configuration first: changing a wait value will not fix a missing routing tag, excluded channel or failed provider delivery.

## Distinguish a wait from a pause

**Sleep mode** pauses the AI when a team member sends a message. **Sleep duration** is measured in minutes. GHL workflow messages also trigger it when **Sleep on workflow messages** is enabled. New contact replies do not extend the existing sleep window.

If there is no reply, check whether the conversation is sleeping, the contact has the exact `ai_off` tag, the contact matches an assistant's active routing tag, and the channel is allowed in Chat Settings. A contact in a single-assistant account still needs the relevant tag match for GHL routing.

## Confirm the message reached the customer

Inspect the conversation and the receiving phone or inbox. A generated response or an “AI Replying” tag is not a delivery receipt. If the provider rejects the message, preserve its error and check the sender/channel configuration.

For support, include the account, contact, channel, message timestamps, expected assistant, and whether a human or workflow recently sent a message. See [common issues](/v3/guides/troubleshooting/first-call-or-chat) and [GHL connection recovery](/v3/guides/troubleshooting/ghl-connection).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.