How to Auto-Triage & Tag Gorgias Tickets with AI
Every support queue is really two problems stacked on top of each other: answering tickets, and knowing which ticket to answer first. Auto-triage solves the second one. When a ticket lands, something has to read it, decide whether it's a where-is-my-order question or an angry chargeback threat, tag it, set its priority, and drop it in front of the right agent — before a human has spent a single minute sorting. Gorgias ships real tools for this, from classic keyword Rules to a newer AI layer that autofills tags and fields. This guide walks through setting up auto-triage and tagging step by step, shows where native Rules run out of road, and stays honest about where an AI agent that reads intent picks up the slack.
What "triage" actually means in Gorgias
Triage is three decisions made the moment a ticket arrives. Categorize it (what is this about — WISMO, return, subscription, wholesale, complaint?). Prioritize it (how urgent — an outage-level VIP escalation versus a routine tracking question?). And route it (who should own it — the returns team, a subscriptions specialist, a senior agent?). Tags are the connective tissue that makes all three work: once a ticket carries a returns tag or a vip tag, you can build views, trigger rules, and report on volume by topic.
Gorgias gives you two engines to do this. The older, deterministic one is Rules. The newer one is the AI layer inside Automate/AI Agent. They're good at different things, and the smartest setups use both — plus, increasingly, a dedicated AI agent layer on top for the reasoning-heavy cases. Let's build it up in that order.
Step 1: Auto-tag the obvious stuff with Rules
Start with what's cheap and deterministic. A Gorgias Rule is an "if this, then that" automation. Per Gorgias' Rules product page, a rule combines a trigger (almost always "when a ticket is created"), conditions (message body text, sentiment, purchase amount, order history, and more), and actions (auto-reply, auto-tag, auto-assign to an agent, auto-close spam).
To auto-tag by keyword:
- Go to Settings → Rules → Add rule.
- Set the trigger to Ticket created.
- Add a condition — for example, Message body contains "refund" OR "return" OR "exchange".
- Add the action Add tags →
returns. - Optionally chain a second action: Assign to → Returns team.
- Save and order it. We go deeper on ordering and conditions in Gorgias Rules explained.
This is genuinely useful and you should do it. Rules are fast, free (they're part of your helpdesk plan, not billed per action), and perfectly predictable. For high-volume, unambiguous phrases — "cancel my subscription," "where is my order" — a keyword rule tags correctly nearly every time.
Step 2: Add AI tagging for the tickets keywords miss
Keyword rules break the moment a customer doesn't use your keyword. "I've been waiting two weeks and nothing's shown up" is a WISMO ticket with zero WISMO keywords. This is where Gorgias' native AI earns its place.
According to Gorgias' documentation on letting AI Agent autofill tags and ticket fields, the AI Agent "reads the context of the ticket conversation and uses your ticket field or tag description to decide the most relevant values to add," and it does this "across both email and chat channels." Setup:
- Open AI Agent settings.
- Under AI ticket tagging, choose Tags or Ticket Fields.
- Click Add and pick which tags/fields to enable.
- Write a clear description for each — this is the instruction the AI uses to decide when to apply it.
- Save.
The description is the whole game here: a tag called damaged-item with the description "customer reports the product arrived broken, cracked, or otherwise physically damaged" will catch phrasings a keyword list never would.
There is one important limitation to plan around, straight from the docs: AI Agent only autofills these properties on tickets that it directly handles itself. So the native tagging is tied to the tickets AI Agent is already working — it's not a blanket, intent-based triage pass over every ticket in your inbox regardless of who handles it.
Step 3: Turn tags into routing
Tags are inert until you route on them. Once triage has reliably tagged tickets, build Views and routing Rules on top:
| Tag / signal | Priority | Route to |
|---|---|---|
vip + negative sentiment | Urgent | Senior / escalations |
returns, exchange | Normal | Returns team |
subscription, billing | High | Subscriptions specialist |
wismo | Low | Automation / self-serve first |
wholesale, b2b | Normal | B2B team |
The point isn't the exact rows — it's the shape: a small, stable tag vocabulary drives priority and ownership, so agents open a queue that's already sorted instead of a firehose.
The honest limits — deflection vs resolution vs automation, and where an AI layer picks up
Native Gorgias triage is good, and you should use it. But be clear-eyed about two seams.
First, keyword Rules read words, not intent. They can't tell an urgent "my order never arrived and my event is tomorrow" from a calm "just checking on shipping." They apply the tag you defined for the phrase you guessed — not the priority the ticket actually deserves. The native AI tagging closes part of this gap, but it's scoped to tickets AI Agent already handles.
Second, watch how the AI itself is billed. Gorgias' AI Agent is priced per automated resolution — per Gorgias' own AI Agent pricing page, that's $1.00 per resolution on a monthly contract or $0.90 annually, with overage interactions at $1.50. And per the Gorgias FAQ, each AI resolution also counts as a helpdesk ticket and is billed accordingly — so the same conversation can show up in two meters. That's a fine model for full deflection, but it makes the economics of using AI purely to tag and route (an action, not a resolution) awkward: you're paying resolution-shaped prices for triage-shaped work.
It helps to separate three things that get lumped together as "AI support":
| Concept | What it means | Billed how |
|---|---|---|
| Deflection | Customer self-serves; no agent touch | Often per resolution/interaction |
| Resolution | The ticket is actually closed out | Per resolution (Gorgias AI Agent) |
| Automation | Discrete actions — tag, set priority, route, look up an order | Per action (Macha credits) |
Triage is automation. It's a sequence of small actions — read intent, set priority, add tags, route — and it's most cost-predictable when you pay per action rather than per resolution. This is the seam an AI agent layer fits into, and it's worth understanding the wider category of AI agents for customer service before you pick one.
Where Macha fits: an AI triage agent on top of Gorgias
Macha is an AI agent layer that runs on top of the Gorgias you already use — the connector is live, and it reads and writes the same tickets your Rules and Views already work with. It does not replace Gorgias or your helpdesk; it does the reasoning-heavy triage that keyword Rules can't.
The screenshot below is Macha's Ticket Triage agent from our demo org. Notice what it's configured to do: read the ticket, set priority (urgent / high / normal / low), and add relevant tags — wired to concrete tools like Update Ticket Priority and Update Ticket Tags. That's intent detection driving automatic tagging, prioritization, and routing, rather than a keyword match.
Because it reads intent, the same agent can handle the ecommerce cases keyword rules miss: it recognizes the two-week-silence email as WISMO, flags a subscription cancellation with a refund threat as high priority, tags a "the box was crushed" message as damaged-item, and routes a wholesale inquiry to the B2B team — all before an agent opens the ticket. And because triage often wants live data, the agent can call a custom tool to check an order's actual status and raise priority when a shipment is genuinely late.
On cost, Macha bills per AI action via credits — never "per resolution." A triage pass (read intent, set priority, add a tag, route) is an automation, and it's priced as one; you can see the model on the pricing page. That keeps triage economics clean: you're not paying a resolution price to sort a ticket that a human still ends up answering.
The clean division of labour: keep Gorgias Rules for cheap, deterministic tagging of the obvious phrases; use native AI Agent for deflection on tickets it fully resolves; and layer a Macha triage agent on top for the intent-driven prioritization and routing across your whole inbox. If you want to see the connector in action end-to-end, we cover it in connecting Gorgias to Macha for ticket replies, and the native side in the Gorgias Automate & AI complete guide.
FAQ
Can Gorgias auto-tag tickets without AI? Yes. Gorgias Rules can auto-tag on keyword and condition matches — trigger on "ticket created," add a condition like "message body contains 'return,'" and add the action "Add tags." Rules are part of your helpdesk plan and aren't billed per action, so they're the right first layer for obvious, high-volume phrases.
How does Gorgias' native AI tagging decide which tags to apply? Per Gorgias' docs, the AI Agent reads the conversation context and uses the description you write for each tag or ticket field to pick the most relevant values, across both email and chat. It only autofills on tickets the AI Agent directly handles, so clear descriptions matter and the coverage is scoped to AI-handled tickets.
How is Gorgias AI Agent billed? Per automated resolution — Gorgias lists $1.00 per resolution on monthly contracts, $0.90 annually, and $1.50 for overage interactions. Note that each AI resolution also counts as a billable helpdesk ticket per the Gorgias FAQ, so confirm both meters against your own volume.
How is Macha's triage billed differently? Macha bills per AI action through credits, not per resolution. A triage pass — reading intent, setting priority, adding tags, routing — is counted as automation actions, which keeps the cost of sorting a ticket predictable even when a human still answers it. See the pricing page for the model.
Does Macha replace Gorgias? No. Macha is an AI agent layer that runs on top of the Gorgias you already use via a live connector. It reads and writes the same tickets, tags, and priorities your Rules and Views rely on — Gorgias stays your system of record; Macha adds the intent-based triage on top.
Ready to let intent, not keywords, sort your queue? Start a free trial of Macha and connect it to your Gorgias in minutes.
Add AI agents to your Gorgias
Macha resolves tickets end to end on Gorgias — no migration, no code.
Zendesk
Freshdesk
Gorgias
Front
Shopify
Stripe
Slack
Notion
Google Workspace
Confluence

