A practical guide to Intercom integrations with AgentKit (2025)

There’s a ton of talk about building custom AI agents for customer support, and it makes sense. The dream is an AI that gets your business, knows your policies, and can handle tricky customer questions all on its own. Two names that keep popping up in this space are Intercom, one of the biggest platforms for customer communication, and OpenAI’s AgentKit, a new toolkit for developers who want to build AI agents from scratch.

Putting these two together sounds like a recipe for some seriously powerful, custom-fit support automation that a standard chatbot just can't touch. But what does it actually take to get them to play nice?

This guide will walk you through what Intercom integrations with AgentKit look like in the real world. We'll cover what you can do with them, what their limits are, the hidden costs, and why a simpler, more direct alternative might actually be the smarter move for your team.

What are the key platforms for Intercom integrations with AgentKit?

Before we get into the weeds, let’s quickly introduce the two main technologies we're talking about.

What is Intercom?

You probably already know Intercom. It’s a hugely popular platform that businesses use to manage all their customer conversations, from live chat and help desk tickets to proactive messages. It’s the central hub for customer interactions, which makes it a prime spot for adding some AI smarts.

A screenshot of Intercom's Fin AI chatbot interface, relevant to the discussion of native Intercom AI features before exploring Intercom integrations with AgentKit.

Intercom does have its own AI, Fin, but some teams find themselves wanting more control. They have very specific, multi-step workflows unique to their business and want an AI that can handle them perfectly. That’s usually when they start looking at developer-focused tools like OpenAI's AgentKit. For most, though, the ideal solution is something that just works with their existing setup without a fuss, which is where a tool like eesel AI for Intercom really comes in handy.

What is OpenAI's AgentKit?

AgentKit isn't a ready-made chatbot. It's a toolkit from OpenAI that gives developers the raw materials to build, test, and launch their own AI agents. Think of it as a low-level framework, not a plug-and-play app. Its main parts are:

• Agent Builder: A visual board where developers map out the logic and flow of an agent’s actions.

• ChatKit: A collection of chat interface components for building custom chat windows.

• Evals for Agents: Tools for developers to test how well their agent is performing against specific data.

• Connector Registry: A system for securely connecting the agent to other tools and APIs.

Basically, AgentKit provides the building blocks for developers to create agents that can handle complex jobs and talk to other services.

How Intercom integrations with AgentKit work

Let’s be upfront: connecting Intercom and AgentKit is not a one-click job. It’s a project that requires a developer to get the two systems talking to each other.

The magic happens through "Tools" in AgentKit's Agent Builder. These tools are just functions that let the AI agent call out to other services, like the Intercom API. This is often done using something called the Model Context Protocol (MCP), an open standard that helps AI models securely use external apps and data. A developer on your team would need to build custom API connectors so AgentKit can do things in Intercom, like send a message or update a ticket.

A typical workflow would look something like this:

1. A customer messages you in the Intercom chat.

2. A webhook from Intercom kicks off a process in AgentKit.

3. The Agent Builder you designed processes the message.

4. If the agent needs info from Intercom (like the user's account details), it uses a connector to ping the Intercom API.

5. The Intercom API sends the data back.

6. The agent figures out a reply and uses the API again to send it back to the customer through Intercom.

The bottom line is you need a developer to write and look after the "glue code" that holds all of this together, including things like API keys and what to do when things go wrong. It’s a powerful setup, but it also means you're completely reliant on your engineering team.

Use cases and benefits of Intercom integrations with AgentKit

If you have the developers to build and maintain Intercom integrations with AgentKit, you can create some really impressive, tailor-made automations that are miles ahead of what an off-the-shelf bot can do.

Here are a few things that become possible:

• Proactive customer check-ins: You could have an agent that watches for user activity in your app (through events sent to Intercom) and automatically starts a chat to offer help if it looks like someone is getting stuck.

• Smarter ticket sorting: Imagine an agent that can understand a highly technical problem, check the user's payment status in Stripe, confirm their account level, and then send the ticket to the right engineer with a neat summary.

• Step-by-step problem solving: An agent could walk a user through a complicated issue, like a refund request. It could ask for the info it needs, check details with other APIs, and only bring in a human once all the basic fact-finding is done.

But that control comes with a catch. Every custom step, every rule, and every API call has to be built, tested, and maintained by hand. And that brings us to the not-so-obvious downsides.

The hidden challenges: Limitations, cost, and complexity

While the possibilities sound great, the day-to-day reality of building a custom integration with AgentKit comes with some big hurdles that can catch businesses by surprise. This is where the DIY dream can start to fade.

You’re tied to your developers

AgentKit is a tool for developers, period. Building and maintaining these integrations isn’t something a support lead can do in their spare time; it needs ongoing engineering resources. Every time you want to tweak the workflow or the Intercom API gets an update, it becomes a task for a developer.

This is a completely different world from tools like eesel AI, which offers a one-click Intercom integration that you can get running in a few minutes without touching any code.

The price tag is a moving target

AgentKit will likely have a consumption-based pricing model, meaning you pay for what you use (per token, per action). Your costs are tied directly to your support volume. If you have a busy month, you could get a much bigger bill than you expected, making it really tough to budget your support costs.

This is where platforms with straightforward pricing are a relief. With eesel AI, you pay a flat, predictable monthly fee. There are no extra charges per ticket resolved, so you're not punished for being successful with automation.

No safety net for testing

AgentKit has "Evals for Agents," but it’s a framework for developers to measure performance in a lab-like setting. It doesn't have a simple, easy-to-use simulation mode where a non-technical person can see how the agent would have handled real past conversations before unleashing it on live customers.

That's a huge plus for eesel AI. Its simulation mode lets you test your setup on thousands of your past tickets. You get a solid forecast of how many tickets it will resolve, so you can feel confident before you flip the switch.

Juggling all your knowledge sources

With AgentKit, it's on you to connect every single place your agent needs to learn from. That means manually setting up file uploads, dealing with vector stores, and building custom connectors to your help docs. It's a lot of work upfront and can become a real headache to maintain.

In contrast, eesel AI is built to connect to your knowledge sources instantly. It automatically learns from your old Intercom tickets to pick up your brand voice and common answers. It also plugs right into places like Google Docs, Confluence, and Notion in just a few clicks.

A simpler path to powerful Intercom automation: eesel AI

For most teams, the time, cost, and general hassle of a DIY solution just aren't worth it. eesel AI delivers the power of a custom AI agent without forcing you to build one from the ground up. It’s designed to solve the exact problems the AgentKit approach creates.

Get up and running in minutes, not months

With a self-serve setup and a one-click Intercom integration, you can have eesel AI working in less time than it takes to finish a cup of coffee. No mandatory sales calls or long onboarding required. You can just sign up and get going.

Full control, no code needed

eesel AI gives you the fine-grained control you'd expect from a custom build, but through a simple interface. You can use the prompt editor to shape your AI's personality and tone, and tell it exactly what it can and can't do. You also get to choose which kinds of tickets get automated, so you're always in the driver's seat.

All your knowledge, all in one place

Forget about building knowledge bases from scratch. eesel AI learns from your past Intercom conversations and connects to all your other help docs with a few clicks. It pulls all your scattered information together into one smart brain for your AI agent.

Test without the risk

The simulation mode lets you see exactly how eesel AI would have handled your past tickets. It's a stress-free way to tweak your settings and roll out automation knowing it’s going to perform well.

Clear, predictable pricing

With eesel AI, you pay a flat monthly fee. No charges per token, no fees per resolution, and no surprise invoices. It’s simple, predictable, and grows with your business, not your ticket count.

Here’s a quick side-by-side look:

Final thoughts on Intercom integrations with AgentKit

At the end of the day, Intercom integrations with AgentKit are on the cutting edge for teams with serious developer resources and a need for a truly one-of-a-kind workflow. If you have engineers ready to build and maintain a custom solution, it gives you ultimate flexibility.

However, for most businesses, the complexity, unpredictable costs, and ongoing maintenance just aren't practical. Platforms like eesel AI offer a much more straightforward and cost-effective way to get top-tier support automation in Intercom. You get all the perks of a custom-trained agent without the engineering headaches.

Ready to automate your Intercom support without the engineering overhead?

Get the power of a custom AI agent with the simplicity of a one-click setup. Start your free eesel AI trial today.