Steer Ultimate: Always-On Auto Replies
⚙️✅ Access: Always-On Auto Replies is part of Steer Ultimate. Upgrade now to get access, click here.

Always-On Auto Replies is different than Steer Ultimate's After-Hours/Custom Hours Auto Replies. Always-On stays active during working hours, acting like a text receptionist that can help decide whether a message can be handled automatically or needs a human touch.
Jump to:
How to Enable Always-On Auto Replies
How to Customize Always-On Intent Behaviors and Responses
How Always-On Messages Appear in the Steer Inbox
ℹ️ Note: For information on Always-On Auto Replies' built-in guardrails to prevent customer attempts to override the tool with malicious prompts, Click Here.
How to Enable Always-On Auto Replies
In Steer,
-
Go to Settings → Communications → Auto-replies
-
Select the Always-On Option
- Select your AI Reply style
- Customize your Intents
- To learn more about customizing Always-On Intents, Click Here
- Click Save.
Note: Right now, you cannot use after-hours/custom and always-on at the same time, you must choose one mode or the other.
How to Customize Always-On Intent Behaviors and Responses
While in the Always-On Auto Replies settings area, you can customize the intents that control how the AI responds to your customers. There are 20 total intents you can customize.

To customize intents:
-
Click the intent to expand the settings menu.
-
Customize the example reply.
- Action Needed: Choose whether the AI should escalate the message to a human or handle the message itself.
- Path A: Action Needed
- The message is flagged for a service advisor to review.
- Alerts are sent via in-app, SMS, and email.
- The conversation is marked Action Needed in the Steer Inbox
- Path B: No Action Needed
-
If the intent is allowed to be handled automatically, the AI responds on its own.
-
The AI can also send links such as: booking and review links.
-
- Path A: Action Needed
- Prioritize intents by using drag and drop.
- The intent list is priority-based. Higher items in the list take precedence over lower ones. For example, if cancel appointment is above service done elsewhere, the cancel intent wins when both could apply. This helps the AI choose the most important behavior when a message matches multiple intents.
- Save your changes.
Note: Toggle OFF any intents you do not want to utilize for Always-On Auto Replies, this will prevent the AI from using the intent to handle messages.
How Always-On Messages Appear in the Steer Inbox
All replies sent by the Always-On AI Agent will show in your Steer Inbox; whether they require action or not. You can filter messages to see only "Action Needed" conversations to prioritize for your team.

Opening a message where the thread is marked as "Action Needed" shows:
-
the AI’s reply
-
a short summary of what the advisor should do next
-
Example: the system identifies a wrong number and tells the advisor to determine the correct contact information.

If a service advisor wants to take over personally, they can pause auto-replies for that thread. Pause options include:
-
30 minutes
-
3 hours
-
1 day
-
1 week
-
Forever
Note: After the pause expires, the AI resumes automatically on the next new message unless paused forever.
How Always-On Auto Replies Handles Malicious Prompting
Always-On Auto Replies has built-in guardrails designed to strictly follow system instructions and ignore attempts to override the system by injecting malicious prompts. This means, the tool should not disclose sensitive shop information, internal instructions, or other restricted details due to someone trying to trick it into doing so.
If a prompt looks suspicious, or tries to steer the Auto Reply agent away from its intended behavior, the system is designed to keep following the approved instructions instead of the injected request. The tool is built to stay on task, protect restricted information, and avoid acting on malicious prompt attempts.
Overall, Always-On Auto Reply is designed to be resilient against injected manipulation and to keep responses aligned with the intended shop-safe behavior.