What this screen does
The AI answering email tickets screen is where you configure how Rep AI handles your incoming email support tickets — when it replies, what it sounds like, what it must never touch, and how it identifies itself to your customers. Everything that controls your email AI lives on this one page.
The screen subtitle says it plainly: "Reduce support workload by letting AI automatically respond to incoming email tickets."
Saving your changes
A Save Changes button sits at the top right of the page. Any setting you adjust on the screen stays unsaved until you click it, so you can change multiple things together and apply them in one action.
Status banner
A banner at the top of the page reflects the current state of your AI email answering, based on your Omnichannel AI plan and usage:
Ready to activate — AI email answering runs on an Omnichannel AI subscription. The banner links straight to the plan so you can switch it on.
Auto-send enabled — "AI is automatically sending responses using your Omnichannel AI ticket package."
Past your package — your package is used up for this cycle. With overage on, AI keeps sending replies and each resolved conversation beyond your package is billed at $1.20. With overage off, AI email answering pauses until you upgrade, renew, or turn overage on — subscriptions that began before overage billing was introduced keep AI drafting replies instead. A CTA takes you to your package.
For details on packages, tickets, and how billing works, see the Omnichannel AI article. For the overage toggle, the charges, and what happens when you change or cancel your package, see Keep Your AI Answering When You Pass Your Omnichannel AI Plan.
Reviewing answers before customers see them
Once AI email answering is active, the AI writes and sends the reply itself, and each resolved conversation consumes one ticket from your Omnichannel AI package.
To see how the AI would handle your email tickets before or alongside that, use Test & Train (AI Training → Test & Train). You can run your email channel through it as often as you like, free, at any time — it never sends anything to a customer and never consumes tickets.
Identify as an AI agent
A checkbox that controls whether the AI's replies disclose that they were written by AI. Leave it checked if you want transparency with your customers; uncheck it if you prefer the AI to reply silently as part of your team.
The agent (user) the AI would answer on their behalf
A dropdown listing the agents in your helpdesk. The selected user is the one whose name and email signature the AI uses when it replies. Pick a real teammate or a generic support agent — whichever fits your brand.
Use specific AI instructions for emails
A toggle that controls whether email replies follow a dedicated set of instructions instead of your general AI instructions. When turned on, your Email AI Instructions apply to email responses only — everything you've already configured for chat and other channels stays untouched.
Email AI Instructions editor
A rich-text editor where you tell the AI how to write its email replies. You can use bold, underline, links, lists, and more to structure your guidance.
Inside this section you'll find:
Generate for me — produces a starter set of instructions for you, so you don't have to write from a blank page.
Polish prompt — rewrites your instructions to be clearer and more effective without changing their meaning.
Expand (↗ icon) — opens the editor in a larger view for longer instruction sets.
Character counter — shows how much room you've used (up to 200 characters per instruction segment).
Helpful examples of what to write here:
"Use a professional tone, avoid emojis."
"When AI cannot answer an email, send a courteous acknowledgment instead of staying silent."
"Always reply in the language of the original email."
Send acknowledgment messages
A toggle that tells the AI to reply with a friendly acknowledgment whenever it can't fully answer (missing information, unsupported request, etc.) — so the customer never gets ghosted. These acknowledgments respect everything else you've set on this screen: your tone instructions, your agent identity, your AI Instructions.
Use macros for consistent email responses
A toggle that tells the AI to use your helpdesk macros as reference material when writing replies. The AI doesn't paste macros verbatim — it uses them as guidance to keep its phrasing aligned with the brand voice your team already uses.
Do Not Answer Rules
A toggle that opens a list of email categories the AI must never reply to. Use this to block sensitive topics — legal escalations, refund disputes, anything you want a human to handle every time. When the AI sees a matching email, it skips it entirely and leaves it for your team.
Answer all tickets
A toggle that decides which tickets the AI is allowed to touch:
Off — the AI answers only tickets that came in by email.
On — the AI also answers tickets your helpdesk created from chat-widget conversations (for example, contact-form submissions that escalated).
Turn this on if your helpdesk is your single source of truth for all support tickets and you want the AI to help across the board.
Stopping the AI on a specific thread
A note at the bottom of the page explains: to stop the AI from answering one particular conversation thread, tag the conversation with the word "Human" in your helpdesk. Once that tag is on the thread, the AI hands the conversation off and stays out of it.
What the AI knows about your customer
Beyond the subject and message history, the AI reads the custom fields on each ticket — the extra information your helpdesk stores alongside a conversation, such as subscription tier, VIP status, order number, account type, or warranty status. These are the same fields your human agents rely on, set either manually by your team or automatically by your helpdesk's rules.
This works automatically — there's nothing to switch on. Whenever a ticket carries custom fields, the AI uses them to write a more accurate, more personal reply:
It reads the fields along with the message, so it doesn't reply with only half the picture.
Empty fields are skipped, so only meaningful information reaches the AI.
The AI weaves the data in naturally — it won't paste raw "field: value" pairs into the reply, and it won't invent values that aren't there.
Sensitive-looking values aren't echoed back to the customer.
Because the AI can now see these fields, your Email AI Instructions can reference them directly — for example: "If the Subscription Tier field is 'Gold', always offer priority shipping." The same applies to Do Not Answer Rules: you can tell the AI to skip any ticket where a field has a certain value, such as a "Ticket Type" of Legal.
Custom field reading works across your connected support helpdesks — Zendesk, Gorgias, Freshdesk, and Kustomer.
The AI also reads the customer's own support history. Before it replies, it looks at what that customer had resolved with you over the last 90 days — up to their ten most recent tickets — so a returning customer gets an answer that picks up where they left off rather than one that starts from scratch. It works the way a good agent works: check the customer's file first, then reply.
This also works automatically, with nothing to switch on, on Zendesk, Gorgias, Freshdesk, and Kustomer:
The AI refers to past interactions in plain language — "I see you reached out about the damaged item last month" — and never quotes ticket numbers or reads internal records back to the customer.
When the customer's history has nothing to do with their new question, the AI leaves it out.
Anything the shopper told your AI in chat is combined with their support history, so the AI answers with the same full picture your own team would have.
Only resolved tickets count, so the AI works from settled outcomes. A ticket still in progress isn't folded into the history behind a separate email.
Support history never holds up a reply. If it can't be assembled in time, the AI answers on the message alone and uses the history from that customer's next email onward.
Rep AI refreshes its picture of a customer about once a week, so tickets resolved since the last refresh get included.
A first-time customer with no history is answered exactly as they are now — nothing changes for them.
Email Answering Analytics
A card at the bottom of the screen with a View analytics link. Opens the analytics view where you can see how the AI has been performing on email tickets — how many it answered, how many it handed to your team, and how that maps to support workload saved.
Putting it all together — a typical first-time setup
Pick the agent the AI will answer as.
Decide whether to Identify as an AI agent or stay anonymous.
Preview the AI on your email channel in Test & Train (AI Training → Test & Train) until the answers read the way you want.
Turn on Use specific AI instructions for emails and click Generate for me to seed a starting point. Tweak the wording with Polish prompt.
Decide whether to enable Send acknowledgment messages and Use macros.
If you have topics the AI must never touch, enable Do Not Answer Rules and add them.
Click Save Changes.
Watch the first answers land in your helpdesk, and adjust your Email AI Instructions from there.
Common questions
Do I need an Omnichannel AI subscription for AI email answering?
Yes. AI answering on your connected helpdesk runs on an Omnichannel AI ticket package — one ticket per resolved conversation. You can subscribe from the Console's billing page and the switch is immediate. Previewing AI answers in Test & Train is free and needs no subscription.
What's the difference between "Identify as an AI agent" and the agent dropdown below it?
The dropdown picks who the AI signs replies as (which user's name and signature appear on outgoing email). The checkbox decides whether to disclose that the reply was written by AI. They work independently — you can sign as a real teammate but still disclose AI authorship, or vice versa.
Will Email AI Instructions change how the AI replies in chat?
No. When Use specific AI instructions for emails is on, those instructions apply only to email replies. Your chat-widget AI keeps following your general AI instructions.
How do I stop the AI from replying to one specific email thread?
Tag the conversation in your helpdesk with the word "Human". The AI will recognize the tag and back off that thread.
Does the AI use the custom fields on my tickets?
Yes. The AI automatically reads the custom fields on each ticket — subscription tier, order number, VIP status, and so on — and uses them to write a more relevant reply. There's nothing to set up. If the AI isn't using a field, it's usually because that field is empty on the ticket (empty fields are skipped) or the value looks sensitive. To have the AI actively act on a field, add a line to your Email AI Instructions telling it what to do with that value.
Why didn't the AI mention a ticket my customer clearly wrote in about before?
The most likely reason is that the ticket is still open — the AI works from resolved tickets only. It can also happen if the ticket falls outside the last 90 days, or if the customer has more than ten more-recent tickets.
Where do I see results?
Click View analytics at the bottom of the screen to see deflection performance for email tickets.