How to Connect Gorgias to Macha and Automate Ticket Replies

Macha is not a Gorgias replacement. It sits on top of Gorgias (and Zendesk, Freshdesk, and Front), reads the ticket, decides what to do, and takes real actions back inside Gorgias through 15 native tools. This guide walks the whole setup end to end: connect with an API key, wire up the trigger that fires on new tickets, and build the agent that writes the reply. It takes about ten minutes.

Macha's Gorgias connector shipped on June 23, 2026. Here's how to use it.

What you'll build

By the end you'll have an AI agent that:

1. Wakes up the moment a ticket is created in Gorgias (or a customer sends a new message).
2. Reads the full ticket thread, the customer record, and your knowledge sources.
3. Drafts a reply grounded in your help docs — and either posts it as a public reply, leaves it as an internal note for an agent to approve, or applies a macro.
4. Updates the ticket — sets status, priority, tags, or routes it to the right team.

You decide how far it goes. A cautious rollout drafts internal notes for human approval; a confident one sends public replies on the question types you trust.

Step 1 — Connect Gorgias with an API key

In Macha, go to Connectors → Support and pick Gorgias. Authentication is by API key, so you'll create one in Gorgias first.

Get your Gorgias API key

1. In Gorgias, open Settings → REST API (you'll need an admin or account-owner role).
2. Note your domain (yourstore.gorgias.com), your username (the email of the account the key belongs to), and generate an API key.
3. Copy the key somewhere safe — Gorgias shows it once.

A note on the account you generate the key from: the agent's replies and notes are attributed to that user. Most teams create a dedicated "Macha" or "AI Agent" user in Gorgias so the automation is clearly labelled in the ticket history and easy to audit later.

Paste it into Macha

Back in the Macha connector screen, enter your Gorgias domain, username, and API key, then save. Macha validates the credentials and the connector goes live.

Once connected, your agents can use all 15 Gorgias tools:

That spread is the point: replying is one action among many. The same agent that drafts a response can read a PDF attachment, find the right macro, set the priority to urgent, and assign the ticket to your returns team — in a single run.

Step 2 — Choose how the agent gets triggered

A connector lets the agent act in Gorgias. A trigger is what starts it. Macha's Gorgias connector gives you four:

• Ticket Created — fires when a new ticket lands. The most common choice for reply automation.
• Ticket Updated — fires on status, priority, assignment, or tag changes.
• New Message Added — fires when a customer replies (or any new message hits the ticket).
• Custom Webhook — Macha generates a webhook URL that you attach to a Gorgias HTTP integration, so you control exactly which tickets fire the agent.

The first three are managed events — pick one and Macha listens. The Custom Webhook is for teams who want surgical control, and it's worth understanding because it's the most powerful option.

Wiring the Custom Webhook (the precise route)

If you only want the agent to run on, say, tickets tagged vip or coming through a specific channel, the webhook route lets Gorgias' own Rules decide. The flow:

1. In Macha, add a Custom Webhook trigger to your agent. Copy the generated webhook URL (and signing secret).
2. In Gorgias, go to Settings → Account → HTTP integration → Manage → Add HTTP integration (Gorgias docs).
3. Set the method to POST and paste Macha's URL. Under Triggers, choose the ticket events you want — Ticket created, Ticket updated, or Ticket message created.
4. Add the auth header Macha gives you (Gorgias supports header auth like Authorization: Bearer <key>), and pass ticket context in the body using Gorgias template variables such as {{ticket.id}} and {{ticket.customer.email}}.
5. Save. Now only the tickets that match the rule conditions you set in Gorgias will reach your agent.

A couple of Gorgias-side realities worth knowing, straight from their docs: HTTP integrations have a 5-second timeout per request and retry up to 3 times with exponential backoff (10s, 20s, 40s). And Gorgias enforces plan-tier rate limits — fire a poorly-scoped rule at thousands of tickets and you'll hit HTTP 429. Scope your trigger conditions tightly; don't point the agent at your entire backlog at once.

Step 3 — Build the agent that writes the reply

With the connector and trigger in place, the agent is the brain. Start from a built-in Gorgias agent template under Connectors (the changelog notes both new helpdesk connectors ship with templates), or build from scratch.

Four things make a reply agent good:

1. Instructions. Tell it plainly what it owns and where its limits are. For example: "You handle order-status, shipping, and returns-policy questions. Read the ticket and the customer record. If the question is answered in the knowledge base, draft a reply in our voice. If it involves a refund over $100, a damaged item, or anything you're unsure about, post an internal note summarising the issue and assign to the Returns team instead of replying."

2. Knowledge sources. Connect your help-centre articles, returns policy, and shipping FAQs as a knowledge source. This is the difference between Macha and a generic LLM: replies are grounded in your documented answers, not invented ones. (Note: the Gorgias connector itself doesn't index Gorgias content as knowledge — you attach your docs separately as knowledge sources.)

3. Tools. Give the agent the Gorgias tools it needs — at minimum Get Ticket, Add Public Reply (or Add Internal Note), and Update Status. For Shopify-native stores, connect Shopify and Stripe too, so the same agent can look up the actual order and payment before it answers "where is my order?"

4. The reply mode. Decide whether the agent sends public replies or drafts internal notes. Start with internal notes on a narrow question type, read what it produces for a week, then promote the categories you trust to auto-send. This is the single most important rollout decision — and the cheapest insurance against a bad day.

Watch-outs: when not to auto-send

Honest guardrails matter more than enthusiasm here.

• Don't auto-send on day one. Run in internal-note (draft) mode first. You'll catch tone issues, edge cases, and knowledge gaps before a customer ever sees them.
• Keep the scope narrow. An agent that "handles everything" handles nothing well. Constrain it to a few well-documented question types and let humans keep the rest.
• Mind the rate limits. Gorgias throttles by plan tier. Scope triggers to the tickets that actually need the agent, not the whole firehose — and never backfill your archive through a live trigger.
• Attribute it clearly. Use a dedicated Gorgias user for the key so every AI action is labelled in ticket history. Your human agents — and your audits — will thank you.
• It's an addition, not a rip-and-replace. If Gorgias' native AI Agent already deflects your simplest FAQs, Macha is the layer for the workflows it can't reach: cross-tool actions (Shopify/Stripe/Slack), your own agent logic, and your own knowledge — all metered as actions, not per resolution. See how the pricing works before you scale.

How Macha's pricing works for this

Macha bills in credits, charged per AI action — not per deflection or per resolution. The default model, GPT-5.4 Mini, is 1 credit per response; models range from 0.5 to 9 credits each depending on which you pick (GPT-5 is 2, GPT-5.4 is 3). A reply agent that reads a ticket and posts one response is a handful of credits; you control cost by choosing the model and scoping the trigger. Start with a 7-day free trial, no credit card required and see the pricing page for plan limits and credit allowances.

This is a different model from Gorgias' native AI Agent, which charges per full resolution. Neither is automatically cheaper — it depends on your resolution rate and how many cross-tool actions each ticket needs. The point of Macha is what the agent can do across your stack, billed transparently as actions.

FAQ

How does Macha authenticate with Gorgias? By API key. You generate a REST API key in Gorgias (Settings → REST API), then enter your domain, username, and key in Macha's Gorgias connector. No OAuth app required.

Can Macha send a public reply automatically, or only draft one? Both. The agent can Add Public Reply (customer-facing) or Add Internal Note (agents only). Most teams start in internal-note mode for review, then promote trusted question types to auto-send.

What can the Gorgias agent actually do besides reply? Fifteen tools: read the full ticket and customer, search tickets/customers/macros, reply or note, update status/priority/tags/custom fields, assign to a user or team, list agents and teams, and read text from attachments (PDF, DOCX, XLSX, CSV, TXT).

Do I need a webhook, or will the built-in triggers do? For most setups, the managed Ticket Created or New Message Added trigger is enough. Use the Custom Webhook only when you want Gorgias' own Rules to decide exactly which tickets reach the agent.

Will this conflict with Gorgias' native AI Agent? No — Macha runs as a layer on top. Scope each to different ticket types if you run both, and attribute Macha's actions to a dedicated Gorgias user so the two are easy to tell apart in ticket history.

Which other tools can the same agent use? Anything you connect: Shopify, Stripe, Slack, Notion, and more. That's the value of the layer — one agent reads the Gorgias ticket and acts across your whole stack.

Get started

Connecting Gorgias to Macha is an API key, a trigger, and an agent — about ten minutes of setup for replies that draw on your real knowledge and act across your real tools. Start a 7-day free trial, no credit card required, open Connectors → Support, and pick Gorgias. The full walkthrough lives in the Macha docs, and you can see every Gorgias tool and trigger on the Gorgias integration page.

Written by Abbas (Customer Support & AI, Macha) · Reviewed by Ankeet Guha (Co-founder & CTO) · Published 2026-06-24 · Last updated 2026-06-24.