Auto-replies on first inbound

An auto-reply is the customer-facing acknowledgement that goes out the moment a new email conversation starts. Ochre supports one workspace-wide auto-reply template for the email channel. It sends once, on the first inbound message of a new conversation, and never again on the same thread. It is off by default.

A bad auto-reply is worse than no auto-reply. "Your ticket has been received and is very important to us" reads like 1998 corporate spam. If you turn the auto-reply on, write it like a person.

Where the template is configured

Settings → Domains — three fields:

• Enabled — toggle the auto-reply on or off.
• Subject — defaults to Re: {subject}. The {subject} placeholder is replaced with the original subject.
• Body — the message itself, plain text or markdown.

There is no separate after-hours template, no per-channel template (email only), no per-routing-rule override, and no Liquid-style placeholder syntax beyond {subject} in the subject line.

When the auto-reply fires

It fires when:

• The conversation is brand new on the email channel (no prior agent messages).
• The inbound message is from a real end user (auto-responders are filtered upstream).
• Auto-reply is enabled in workspace settings.
• The AI agent is not in autopilot/auto-send mode for the email channel. When the AI is responding directly, the auto-reply is suppressed so the customer doesn't get a duplicate "we got it" right before the real answer arrives.

It does not fire on subsequent messages, on conversations an agent created outbound, or on detected auto-responder loops.

Loop protection

Outbound auto-replies are stamped with Auto-Submitted: auto-replied so the recipient's mail server, and Ochre's own inbound webhook filter, won't bounce another auto-reply back at us. The post-ingest pipeline also drops anything carrying the Auto-Submitted header before it reaches the auto-reply path, so two auto-responders can't ping-pong forever.

How the auto-reply shows up

To the customer, it arrives as a normal outbound email from your verified domain, threaded against their original message ID. It looks like a fast first response, except it's instant and obviously templated.

In the inbox, the auto-reply appears as an outbound message in the conversation thread with author_name set to the workspace name. Agents see it inline so they don't accidentally repeat the wording in their first real reply.

What the auto-reply does not do

• It is not a substitute for a real response. The SLA clock keeps ticking; the customer is still waiting on a teammate.
• It does not send on the live chat widget or on Slack channels. Those have their own immediacy expectations.
• It does not fire when the AI is auto-sending in that conversation — the AI's reply is the response.

When to turn it on

The honest case for an auto-reply is when:

• Replies routinely take longer than a customer expects.
• You don't have autopilot enabled, so a human is the first responder.
• You want to set expectations explicitly ("we're back at 9am").

If the AI is replying directly within seconds, you don't need an auto-reply.

Related

• Routing overview for the bigger picture.
• Custom domain email for the domain the reply goes out from.
• Auto-close for the closing-up template at the other end of the lifecycle.