Whether you're building a virtual assistant, an onboarding guide, or a powerful support agent, Cogfy Chat gives you the tools to create conversational experiences tailored to your needs. At its core, the chat feature is fully modular — powered by a flexible system of middlewares that let you define how your assistant behaves, responds, and interacts with users.

With just a few clicks, you can combine powerful components like history tracking, custom instructions, and leading AI models to shape your assistant’s personality and intelligence. Want to go further? Add voice capabilities, use tools to automate workflows, or dynamically change behavior with field-based conditions.

This post will walk you through the foundations of Cogfy Chat, so you can start building smart, adaptive conversations right inside your product.

What is Cogfy Chat?

Cogfy Chat is a flexible and powerful way to bring conversational AI into your web app—without writing a single line of backend code. By adding a chat field to any collection, you give users the ability to interact with an AI agent that understands the context of the record it's tied to. This makes it perfect for building support bots, onboarding assistants, appointment schedulers, or even internal tools that guide your team through complex workflows.

Each conversation lives inside a collection record, so chats are not just floating sessions—they're structured and stored, with access to any other field in the record. And the best part? The chatbot’s behavior is entirely defined by you, through a set of modular features called middlewares. With a few configuration steps, you can shape a chatbot that fits your business and adapts to your users.

In the next section, we'll dive into how this middleware system works and how it gives you full control over the chat experience.

How It Works: The Middleware System

When you create a chat in Cogfy, what you’re actually building is a stack of middlewares—modular features that run in sequence to control how the chat behaves. Think of them as building blocks: each one does a specific job, and together they define how the AI thinks, responds, and interacts with your users.

There are three core middlewares you’ll see in almost every chat setup:

• History: This one sends the full conversation history to the AI model so it has context about what’s been said before. Without it, the LLM would treat each message like a blank slate.
• Instructions: This middleware gives the model its role or purpose. It’s like setting the AI’s job description. You can use it to define tone, goals, restrictions—anything you want the assistant to keep in mind.
• Model: This defines which large language model will be used for the chat. You can choose between providers like OpenAI, DeepSeek, Grok, or open models like LLaMA. Each has its own strengths, and you can swap them out depending on your needs.

That’s the basic engine. Just three middlewares and you already have a fully functional AI chat running inside your app—tuned to your tone and logic.

Enhancing the Chat: Powerful Middleware Additions

Once you've got the basics working—conversation history, system instructions, and your model of choice—you can start supercharging your chat by layering in additional middlewares. These aren't required, but they unlock some really compelling use cases that can turn your chatbot from a cool feature into a real productivity booster.

Here are some of the most useful additions:

• Transcribe Audio: Want your users to speak instead of type? This middleware converts voice messages into text before they hit the LLM. It’s great for accessibility, mobile experiences, or anytime voice feels more natural.
• Generate Speech: Flip the experience around—this middleware turns the model's response into audio using TTS (text-to-speech). Add it and you've got a fully voiced assistant, perfect for hands-free or embedded experiences.
• Tool: This is where things get really powerful. The Tool middleware lets the LLM call external functions—like hitting an API, saving data to your collections, or even triggering workflows. It turns your chatbot into an actual operator, not just a text responder. (We’ll go deep on this in the next blog post.)

You can mix and match these middlewares depending on what you want your chat to do. Want a voice-controlled booking assistant? Add Transcribe Audio, Generate Speech, and Tool. Need a quiet in-app helper that just reads the docs and answers questions? Stick to the core three.

These additions are what make Cogfy chat more than just a text box. They're how you adapt the chat to your app, your product, and your users.

Setting Up Your First Chat

Ready to get hands-on? In this video, we’ll walk you through creating your first chat in Cogfy—from adding the chat field to configuring the core middlewares. It’s a quick setup, and once you’ve seen it done, you’ll realize how flexible (and fun) it is to build your own chatbot experience.

👉 Watch the video below to see the full setup in action:

[Vídeo de setup básico]

Next Steps

Now that you’ve seen how to set up a Cogfy chat and connect it to a language model, you’re already equipped to build powerful conversational experiences. But the real magic happens when your chat goes beyond conversation—when it starts doing things.

That’s where Tools come in.

In the next post, you’ll learn how to give your chatbot superpowers: creating records, calling APIs, triggering workflows, and more—just by talking. Think of it as turning your chat into an interface for action.

Explore it next and take your Cogfy chat to the next level.