A practical overview of Intercom's Fin Guidance

AI agents are popping up everywhere in customer support, but let's be honest, their real value comes down to one thing: control. An AI left to its own devices can cause more headaches than it solves, leaving customers confused and your support team with a mess to clean up. Intercom, a big name in the customer service space, created a feature called Fin Guidance to help teams rein in their AI agent, Fin. The promise is that you can align the AI’s responses with your brand voice and support policies, making sure your bot doesn't go rogue.

In this guide, we're going to take a practical look at Fin Guidance. We'll break down what it is, its main features, and, more importantly, where it falls short when it comes to scaling, maintenance, and cost. Let’s dig in and figure out if a system built on manual rules is really the best way to manage a modern support AI.

What is Intercom's Fin Guidance?

Fin Guidance is a feature inside Intercom that lets you direct its AI agent, Fin, using plain English instructions. You can think of it as a playbook or a set of guardrails for your bot. Instead of writing code, you just write simple commands to tweak how Fin talks to customers.

According to Intercom, the main goals here are to:

• Shape its personality: You can define the AI's tone and style to make sure it sounds like a genuine part of your team.

• Direct its actions: You can tell Fin what to do in certain situations, like when a customer brings up a competitor or asks about a sensitive topic.

• Control escalations: It lets you create clear rules for when a conversation needs to be passed to a human, stopping the AI from tackling problems it isn't ready for.

• Prioritize knowledge: You can point Fin to specific help center articles or content you want it to use when answering certain questions.

Basically, Fin Guidance works by building up a library of explicit rules that the AI checks before it answers. This gives you a decent amount of manual control over how the bot behaves.

Key features of Fin Guidance

Fin Guidance gives you a lot of detailed control, which can be useful if you have the time and people to manage it all. It’s all about creating specific instructions for different scenarios.

Controlling behavior with Fin Guidance categories

Intercom splits its guidance into a few main categories so you can build out your AI's rulebook one piece at a time.

1. Communication style: This is where you set the AI's general vibe. You can specify certain words to use (e.g., "always call our products ‘plans,’ not ‘subscriptions’"), limit jargon, and even tell Fin to use emojis to keep things friendly.

2. Context and clarification: You can teach Fin to ask follow-up questions instead of just guessing. For example, you could set a rule like, "If someone reports a bug, ask for their device and operating system before doing anything else." This helps cut down on the back-and-forth.

3. Handover and escalation: This one is a big deal for managing risk. You can make rules that automatically send a chat to a human agent if sensitive words like "cancel account" or "billing error" pop up. This makes sure tricky issues always get a human touch.

4. Content and sources: If you have one perfect article on a topic, you can force Fin to use it. For instance, a rule could be, "For any questions about refunds, you must use the info from the 'Official Refund Policy' article."

Personalizing responses with Fin Guidance audience rules

On top of the general rules, Fin Guidance lets you personalize instructions for specific users.

• Audience targeting: You can apply rules to certain customer groups. For example, you could give special, high-touch instructions for your "VIP Customers" or provide different answers to users on a "Pro" plan versus a "Free" plan.

• Attributes: You can also pull in customer data from Intercom, like their name or plan type, to make the bot's messages feel a bit more personal.

While this sounds good, it all hinges on having perfectly clean and updated data in Intercom. If that data is missing or wrong, the personalization just breaks. A more reliable way to personalize is to let the AI learn from the full history of past support conversations. For instance, platforms like eesel AI learn the nuances from thousands of past tickets to understand a customer's history and what they actually want, rather than just relying on a few predefined data fields.

The operational reality: Limitations of Fin Guidance

While a rule-based system like Fin Guidance seems perfect on paper, it can create some serious operational headaches and doesn't always grow with your company. Here are some of the real-world challenges you're likely to run into.

Fin Guidance: High maintenance and conflicting rules

Even Intercom suggests creating separate, focused guidance for each goal and being careful not to create contradictions. That’s much easier said than done. As your business changes, you’ll keep adding more and more rules. What happens when a new rule for "Pro" plan customers clashes with an older rule about billing? You often get unpredictable AI behavior, where Fin either gives a weird answer or just gets stuck.

This turns into a constant maintenance chore. Your team has to manually go through and untangle a growing web of rules just to make sure the AI works right. It’s a reactive cycle that can easily eat up someone's entire job.

This is a problem that more modern AI platforms handle differently. For example, eesel AI has a simulation mode. Instead of just spot-checking rules one by one, you can test your AI setup against thousands of your actual past support tickets in a safe environment. You get a solid forecast of its resolution rate and can see exactly how it would have answered real customer questions before you ever turn it on for live customers. This lets you build confidence and catch problems early, without all the guesswork.

The eesel AI simulation dashboard shows how AI performance is tested against past tickets before deployment, a key difference from Fin Guidance.

The need for a perfect, static knowledge base

The "Content and Sources" feature in Fin Guidance completely relies on you having a perfectly organized and up-to-date knowledge base. The AI is only as smart as the articles you manually tell it to read. If an article is out of date, or if the best answer is buried somewhere else, the AI is going to stumble.

This approach doesn't really jive with the reality that useful information is often scattered all over the place. What about those detailed troubleshooting steps in a Google Doc, the latest product updates in Confluence, or the clever solutions shared in past support tickets? Fin Guidance can’t get to that information without a ton of manual setup.

A better way is to connect all your knowledge from day one. An AI agent from eesel AI can plug directly into your help desk, internal wikis like Confluence and Google Docs, and even learn from the resolutions in your past tickets. This lets the AI find the best answer on its own, no matter where it is, without needing you to write dozens of fragile rules pointing to specific articles.

This infographic shows how eesel AI connects to multiple knowledge sources, unlike the more limited approach of Fin Guidance.

Feature	Intercom's Fin Guidance	eesel AI
Control Method	Writing manual, rule-based instructions	Unified prompt editor + self-learning from data
Knowledge Sources	Mostly help center articles (manual linking)	All sources (past tickets, Confluence, GDocs, etc.)
Scalability	Gets complicated as rules grow and conflict	High; AI adapts as your knowledge grows
Pre-launch Confidence	Manual previewing of individual answers	Bulk simulation on thousands of past tickets
Setup Time	Days or weeks to write and test rules	Go live in minutes with one-click setup

The unpredictable cost of Fin Guidance

Let's talk about the one thing every support leader has to worry about: the budget. This is where Intercom's model for Fin can become a real problem.

According to Intercom's own site, their pricing is $0.99 per resolution. This is an extra charge on top of their standard platform fees, which already include a per-seat cost for your human agents.

Here’s what that actually looks like:

• Unpredictable bills: Your costs are directly tied to how many tickets the AI resolves. If you have a busy month and Fin does a great job, your bill will be much higher than you expected.

• A budgeting nightmare: It's almost impossible to guess what your monthly AI costs will be. This makes financial planning tough and can lead to some awkward conversations at the end of the quarter.

• A penalty for success: In a strange way, this model punishes you for automating more tickets. The better your AI performs, the more you pay. This is the exact opposite of why most teams adopt automation in the first place, which is to become more efficient and manage costs.

In contrast, eesel AI's pricing model is designed to be predictable. We offer clear plans based on a set number of monthly AI interactions. There are no per-resolution fees. You know exactly what you'll pay each month, and you're never penalized for your AI doing its job well. Plus, with flexible month-to-month plans, you can get started without getting locked into a long-term contract.

A view of eesel AI's predictable, transparent pricing page, which contrasts with the per-resolution model of Fin Guidance.

Fin Guidance: Moving beyond manual rules for a smarter AI

Fin Guidance gives you a very hands-on way to control Intercom's AI agent. For teams that want to script every little interaction, it provides the tools you need. But that control comes with a hefty price: a ton of ongoing work, a high risk of things getting messy as you scale, and a confusing pricing model that can work against you.

The future of AI in support isn't about writing an endless list of rules. It’s about using an AI that can learn from all of your company’s knowledge, be tested safely before it ever talks to a customer, and operate on a budget you can actually predict. It's about working smarter, not harder.

If you're looking for an AI solution that works with the tools you already use, like Intercom, gives you more powerful and intuitive controls, and puts you back in charge of your budget, it might be time to check out a different approach.

See how you can get started with eesel AI in just a few minutes and run a simulation on your own support tickets today.