Gorgias AI guidance to control handover to human agents: A 2025 overview

Everyone's talking about AI in customer support, and for good reason. The idea of instant answers, 24/7 availability, and a lighter load for your team is pretty appealing. But there’s a catch we don’t always talk about: the handover.

What happens when the AI gets stuck and needs to pass a customer to a human agent? That moment is critical. If the transition is clunky, slow, or just plain wrong, it doesn't just frustrate your customer. It actually creates more work for your team and completely undermines the efficiency you were aiming for in the first place.

The real challenge here is control. How do you teach your AI to know exactly when to solve a ticket on its own and when to raise its hand for help? Getting this balance right is the secret to a great customer experience and a support system that actually works.

This guide will walk you through how Gorgias AI uses its "Guidance" feature and a few other tools to manage that handover process. We’ll look at what it can do, where you might run into some limitations, and explore a more straightforward way to get total control over your AI support workflows.

How Gorgias controls AI-to-human handovers

In the Gorgias world, "Guidance" is a feature that lets you write custom, plain-language instructions for your AI Agent. Think of it as giving your AI a playbook on how to handle specific topics or scenarios. This helps make sure it acts in line with your brand’s policies and sticks to your unique tone of voice.

But here’s the thing: controlling when the AI hands off a ticket to a human isn’t just about the Guidance feature. It’s actually a mix of three different tools you have to use together:

• Guidance: Your detailed instruction manual for specific situations.

• Handover Topics: A simple list of subjects the AI should always pass over to a human.

• Rules: A separate engine that tells the AI to ignore certain tickets entirely.

It helps to think of it like onboarding a new support agent. You’d give them a detailed training manual (that's your Guidance), a clear list of problems they should immediately send to a manager (your Handover Topics), and instructions on which emails to just ignore, like spam (your Rules). When used together, these features are meant to shape the AI Agent into a digital team member that gets how your business operates.

How to control AI handovers in Gorgias

Gorgias gives you a few different levers to pull when managing AI escalations, but you’ll need to jump into different parts of the platform to get everything configured. Let's break down how each piece of the puzzle works.

Using handover topics for predefined escalations

The most direct way to control handovers in Gorgias is with "Handover topics." It’s pretty much exactly what it sounds like: you create a list of subjects that the AI should always pass to a human agent, no questions asked.

This is the right tool for the clear-cut cases where you know a human touch is non-negotiable. Some common examples you might add to this list are:

• Anything related to invoices and billing inquiries

• Conversations with angry or frustrated customers

• Questions about privacy or personal data

• Complex technical issues that need real troubleshooting

It’s a bit of a blunt instrument, sure, but it’s effective for making sure sensitive or high-stakes conversations land with your team right away.

Creating excluded topics with rules

Sometimes you don't want the AI to escalate a ticket, you want it to ignore it completely. For that, Gorgias has you use its standard "Rules" engine, which lives in a totally different area from the AI Agent settings.

The process involves creating a rule that automatically adds an "ai_ignore" tag to tickets that meet certain criteria. For instance, you could set up a rule to ignore any tickets coming from a specific email address (like a partner you work with) or any message that contains the word "giveaway."

While this works, it means a key part of your AI's brain is managed in a separate corner of the helpdesk. It forces you to get comfortable with the Rules engine and, more importantly, to remember to check there whenever you’re trying to figure out why your AI did (or didn’t do) something.

A screenshot of the Gorgias interface, showing how to use the Rules engine for the Gorgias AI guidance to control handover to human agents.

Writing custom instructions with the Guidance feature

The "Guidance" feature is where Gorgias gets its most flexible. This is where you can move beyond simple "always escalate" or "always ignore" commands and start building more nuanced logic for your AI.

Guidance lets you write detailed, step-by-step instructions for the AI in plain English. For example, you could write something like: "If a customer asks about a return but their order is over 30 days old, explain our return policy, state that the window has closed, and do not process a return. Offer them a 10% discount code for a future purchase instead."

Gorgias tells the AI to prioritize these instructions over the general knowledge it pulls from your Help Center, giving them the final say. This is where the real logic for smart handovers lives, but getting it right takes careful writing, constant updates, and a whole lot of testing.

A screenshot of the Gorgias AI Guidance feature, which is central to the Gorgias AI guidance to control handover to human agents.

Limitations of the Gorgias handover process

While Gorgias gives you the tools to manage AI handovers, the way they're put together can create some real headaches for support teams just trying to build a reliable, automated workflow.

A fragmented setup across multiple settings

Honestly, the biggest frustration is that your AI's control logic is scattered all over the platform. Your Handover Topics are in one part of the AI Agent settings, your Exclusions are managed in the main Rules engine, and your detailed Guidance lives in the Knowledge section.

This fragmented approach makes it incredibly difficult to get a single, clear view of your AI's decision-making process. If you're trying to figure out why a ticket was handed over, you literally have to check three different places. This can make troubleshooting and fine-tuning a slow, confusing process, especially as your business rules get more complex. A centralized workflow engine, where all the logic and actions are in one place, just makes more sense.

Limited ability to test and simulate before going live

Gorgias lets you test AI responses one by one in a sandbox. That's fine for checking a single instruction, but it doesn't give you any idea how your changes will actually perform at scale.

There’s no way to run a bulk simulation on your past tickets. You can't see how your newly written Guidance would have handled the 500 tickets you got last week. This means you have to deploy your changes, cross your fingers, and hope for the best. You're stuck reactively watching the "Handover view" to catch mistakes after they've already reached a customer.

Modern platforms like eesel AI have solved this with powerful simulation modes. You can test your entire AI setup on thousands of your real, historical tickets to get an accurate forecast of its performance and cost savings before a single customer ever interacts with it.

A reactive coaching model instead of proactive control

The whole process for improving the Gorgias AI is designed to be reactive. An agent reviews a ticket the AI handled, gives the response a "Good" or "Bad" rating, and maybe suggests a better knowledge source for next time.

This "coaching" model means you're always playing catch-up, fixing mistakes after they’ve already happened. It's a necessary step, but it’s a direct result of the lack of proactive testing. Without the ability to simulate changes, teams are forced into a constant cycle of monitoring and coaching, which eats up valuable agent time that could be spent on more important conversations.

Pricing that penalizes you for volume

The Gorgias pricing model can add another layer of stress. AI interactions are often bundled into plans, but they can also count toward your overall billable ticket limit.

Here’s the problem: an AI that isn't perfectly configured and hands over too many tickets can drive up your costs without warning. In many cases, you might end up paying for the initial AI interaction and for the human agent's time to resolve the now-escalated ticket. This makes it tough to predict your monthly bill and can punish you for having an AI that's still learning. In contrast, tools like eesel AI offer plans based on overall capacity with no per-resolution fees, so your costs are completely transparent and predictable from day one.

A better way to control AI handover: eesel AI

The challenges of a scattered, reactive system really highlight the need for a more intuitive and powerful solution. If you're looking for total control over your AI support, here’s how eesel AI offers a much better approach.

• Everything in one place: With eesel AI, all your AI's logic is managed in one simple, self-serve dashboard. Knowledge sources, escalation rules, and custom actions all live together. You won’t have to jump between three different sections of your help desk just to understand how your AI is going to behave.

• Total control and a gradual rollout: eesel AI’s workflow engine gives you fine-grained control over what gets automated. You can start small by telling the AI to only handle simple "where is my order?" tickets and escalate everything else. As you get more comfortable, you can gradually expand its scope without any risk.

• Test with confidence: The simulation mode is a huge advantage. Before you even think about going live, eesel AI can run your setup against thousands of your past tickets. You get a precise report showing what the automation rate would have been, which tickets would have been resolved, and where the gaps in your knowledge are. No more guessing.

• Go live in minutes: Getting started is ridiculously simple. You can connect your help desk (including Gorgias) to eesel AI with a single click and build your first AI agent on your own in minutes. Forget about sitting through lengthy sales calls and mandatory demos just to see if it works.

Understanding Gorgias AI pricing

To give you the full picture, here’s a breakdown of Gorgias’s pricing plans that include their AI Agent. The model is based on the number of helpdesk tickets, with a certain number of AI interactions included in the higher tiers. Just keep in mind that overages can apply if you exceed these limits, which can lead to those unpredictable costs we talked about earlier.

Plan	Monthly Price	Included Helpdesk Tickets	Included AI Interactions	Overage Cost (per 100 tickets)
Starter	$10/mo	50	0	$0.40 / ticket
Basic	$60/mo	300	60	$40
Pro	$360/mo	2,000	600	$36
Advanced	$900/mo	5,000	2,500	$36

Taking control of your AI handovers

Controlling the AI-to-human handover isn't just a small technical detail; it's fundamental to delivering a great customer experience every single time. While Gorgias offers tools like Guidance, Handover Topics, and Rules, the system's fragmented nature and lack of powerful testing features often leave teams in a reactive mode, dealing with unpredictable costs and behavior.

For teams that need total, proactive control over their support automation, a unified and test-driven platform is the only way to go. You should be able to build, test, and deploy AI with full confidence in how it will perform.

Instead of just hoping your AI works, what if you could know for sure? See how eesel AI lets you simulate, control, and deploy AI with complete confidence. Start your free trial today.