A guide to ChatKit Entity Tagging: What it is and how it works

Building an AI chat experience that’s actually helpful is about more than just getting text-based answers. We've all been there, stuck in a frustrating loop with a bot that just doesn’t get it. For a conversation to feel real, the AI needs to understand the key "things", the people, products, and documents, that make your business tick. It has to connect the dots.

A neat feature for making these richer, more contextual conversations happen is entity tagging. It’s what can turn a simple chat box into an interactive workspace. This guide will give you a clear, no-fluff overview of ChatKit Entity Tagging, explaining what it is, what it can do, and, just as importantly, where it might not be the best fit for businesses that need to get things done.

What is ChatKit Entity Tagging?

Before we get into the tagging part, let's quickly talk about ChatKit itself.

What is OpenAI's ChatKit?

OpenAI's ChatKit is a JavaScript framework that helps developers build the user interface for AI chat. It gives you the front-end components for a polished chat experience, with things like UI customization and real-time response streaming. Think of it as a starter kit for the "chat bubble" part of your app. The main thing to remember is that it’s a front-end tool, not a complete, out-of-the-box solution, which has some pretty big implications for how you build and maintain it.

What is entity tagging?

Entity tagging is basically the process of recognizing and linking specific bits of information, or "entities," right inside a conversation. It’s about making your chat interface aware of the important nouns in your business world.

Instead of just seeing a line of text, the system identifies meaningful things. For a typical business, these entities could be:

• A customer's name

• A support ticket ID ("TICKET-54321")

• A product name or SKU ("SKU-QB-1138")

• A document title ("Q4 Onboarding Policy")

• An order number ("#100-555-1234")

With ChatKit Entity Tagging, developers can let users reference these items directly in the chat, usually by typing an "@" to pull up a searchable list of relevant entities.

Core capabilities of ChatKit Entity Tagging

So, what can you actually do with this feature? Let's look at its main functions and what they mean for creating a smarter chat experience.

Enable @-mentions for interactive context

The most familiar use for entity tagging is enabling @-mentions. Just like tagging a coworker in a document, users can type "@" to search for and tag specific things like other users, documents, or products.

For a developer, this is handled with a function that runs a search every time a user types after the "@" symbol. The business use here is pretty cool. For example, a support agent chatting with a customer could @-mention a specific knowledge base article from Confluence to share a direct link to the solution, all without having to leave the chat window.

Use cited sources for trustworthy answers

Let's be honest, one of the biggest problems with AI is trust. How do you know the AI isn't just making things up? Entity tagging helps with this by letting the AI cite its sources. When an AI assistant answers a question, it can include a direct link to the tagged item it used for reference.

This grounds the AI's responses in your company's actual data, making them much more reliable. An internal sales chatbot could answer a question about a client's contract terms and provide a clickable tag that links straight to the contract stored in Google Docs. This changes the conversation from "I think this is the answer" to "here's the answer, and here's the proof."

Customize hover-previews for a richer UI

Context is everything. Entity tagging in ChatKit lets you create custom pop-up previews that show up when a user hovers over a tagged item. This provides immediate, relevant information without derailing the conversation or making people open a bunch of new tabs.

Imagine a chatbot for an e-commerce store. A customer support agent could tag a a product, and hovering over that tag could instantly show a preview with the product image, price, and stock level, pulling that info directly from a platform like Shopify. It’s a small UI detail that makes a huge difference in how smoothly things run.

The challenges of implementing ChatKit Entity Tagging

While the features sound great, the reality of putting ChatKit Entity Tagging into practice comes with some big hurdles, especially for businesses looking for a practical, scalable solution.

Significant developer work is required

Here’s the catch with ChatKit: it only gives you the UI components. All the important backend logic, how to search for entities, where to get the preview data from, and what happens when an entity is clicked, has to be custom-built by your developers.

This means you’re not just switching on a feature; you’re kicking off a whole new development project. It takes a good amount of engineering time to build and, more importantly, to maintain as your data sources and business needs evolve. For teams that need to launch powerful AI without a dedicated squad of engineers, a platform like eesel AI offers a self-serve experience. You can connect all your knowledge sources and get an AI agent up and running in minutes, not months.

You get locked into the OpenAI ecosystem

ChatKit is an OpenAI product, and it’s built to play nicely within OpenAI's world. As some early feedback from the community has pointed out, it’s not designed to be flexible. If you build your entire chat experience around ChatKit and later decide you want to switch to a different AI model, like Anthropic's Claude, you’re looking at a complicated and expensive migration. This is a real risk for any business that wants to stay in control of its tech stack. You’re not just adopting a UI framework; you’re tying a core part of your user experience to a single vendor.

It doesn't unify your knowledge

ChatKit itself doesn't bring all your knowledge together; it just gives you a window to display it. To make your entities "taggable," you have to build and maintain separate, custom integrations for every single place your knowledge lives. Want to tag info from your helpdesk, your company wiki, and your cloud storage? That’s three different backend projects your team has to build and look after.

This is a world away from an integrated platform like eesel AI, which connects to all your knowledge sources right away. With a few clicks, it can learn from past tickets in Zendesk, articles in your help center, and internal documents in Notion to provide complete context straight out of the box.

Why ChatKit Entity Tagging is so valuable (and why the platform matters)

Getting entity tagging right can make a huge difference for both your customer support and your internal teams. Here's why the functionality itself is so useful:

• Better first contact resolution: When agents and bots can instantly pull up the right customer data, order details, or knowledge articles, they can solve problems faster and more accurately the first time around.

• Shorter agent training time: New agents can get up to speed much quicker when they have a copilot that can instantly reference company policies or pull up solutions from similar past tickets.

• Simpler internal Q&A: Employees get the answers they need without all the back-and-forth. An internal bot in a tool like Slack can directly tag the right person, team, or project document, which cuts down on a lot of wasted time.

But getting this value isn't just about having the UI feature. It's about the platform that powers it. A developer framework like ChatKit gives you the building blocks, but an end-to-end AI platform is what actually gets you the business result.

eesel AI is built for exactly this. Its AI Copilot doesn't just draft replies; it truly understands your business by learning from thousands of past tickets to suggest the most relevant macros or internal documents. And with a simulation mode, you can test how the AI will perform on your historical tickets before it ever talks to a real customer, a key feature for building confidence that you won't find in a basic framework.

Comparing pricing: ChatKit Entity Tagging vs. a unified platform

Money is always part of the conversation, and the difference between a framework and a platform becomes crystal clear when you look at the costs.

The ChatKit cost model (via the OpenAI API)

ChatKit is open-source, so you don't pay a license fee to use the framework itself. The real cost comes from the OpenAI API calls that your custom-built backend has to make for every search, every response, and every action. This is usually priced per token (for both the input you send and the output you get back).

The big downside here is the lack of predictability. If you have a busy month with lots of customer chats, your API bill can shoot up without warning, making it incredibly hard to budget for.

The eesel AI alternative

eesel AI offers a clear, predictable path. Pricing is based on a set number of AI interactions per month, and you don't get charged per resolution. This lets you forecast your costs accurately and scale your support without being penalized for your success.

Feature	ChatKit (via custom OpenAI backend)	eesel AI
Pricing Model	Usage-based (per token)	Flat-rate (per interaction tier)
Cost Predictability	Low (Varies with usage)	High (Fixed monthly/annual cost)
Hidden Costs	Developer time, server hosting, maintenance	None. All-inclusive plans.
Getting Started	Requires custom development	Free trial, go live in minutes.

Moving beyond the ChatKit Entity Tagging framework

ChatKit Entity Tagging is a great concept. It perfectly shows how AI chat can become a much richer, more contextual, and more useful tool for everyone.

But, as a developer-first framework, it comes with real limitations for most businesses. It demands a ton of developer work and ongoing maintenance, locks you into the OpenAI ecosystem, and its pricing model can lead to unpredictable bills.

For most companies, the goal isn't just to build a chat UI; it's to solve real support and knowledge problems without a massive headache. That calls for a complete, integrated platform that works from day one, not just a box of parts.

Ready to launch a powerful AI assistant that unifies your knowledge and works with the tools you already use? See how eesel AI can automate your support and empower your team. You can be live in minutes, not months. Try eesel AI for free today.