AI Product Landscape

Note: I am not endorsing any products here. Opinions are my own.

There are so many products out there and it can be confusing. Understanding which product does what is critical to speak the AI “lingo”.

We first have the LLM model companies that build LLMs. The LLM themselves are not products we directly use, but is the underlying core. See LLM Models.

• OpenAI makes GPT models e.g. GPT 5.5
• Google makes Gemini models e.g. Gemini 3.1 Pro
• Anthropic makes Claude models, e.g. Claude Opus 4.7

Now we get into the products.

AI products are confusing because they increasingly overlap: every tool is trying to do everything. The truth is that many products are really doing very similar things.

So don't trying to memorize which product does exactly what. Pick one major product in each category, then you will realize most products are surprisingly similar.

Lets break them down into 6 categories. Now, this categorization is not exact. Products in one category can often have features in other categories, as we will see.

1. General chat assistants

E.g. ChatGPT, “Claude Chat”, “Gemini Chat”, Perplexity This is your most common tools that use chat/conversation as an interface.

For practical purposes, you can consider these products to be pretty much the same. E.g. Claude Chat is just Anthropic’s ChatGPT. There are some subtle differences in how each product is positioned, but today the difference is often minor for practical usage. E.g. Perplexity is advertised as specializing in web search, which is perhaps true to certain extend. But the same web search capability is also present in e.g. ChatGPT, Claude. It comes down to personal preference when choosing between them (or which tool is approved by your organization).

Note: “Claude Chat” is actually not a thing. I used Claude Chat to make it more clear. The actual name of this product is just “Claude” (https://claude.ai/new). So Claude is the official name for Anthropic’s chat assistant, which can be confusing since the name of the LLM model made by Anthropic is also “Claude”. When people say “ask Claude”, what they probably mean is to ask the Claude chat assistant. Same thing for “Gemini”.

These products cannot directly use files on your computer; you must explicitly upload them to a chat.

2. Computer / local-work agents

E.g. Claude Cowork, OpenClaw. In contrast to chat assistants, this category of products work directly on your computer, which can be more convenient. E.g. claude cowork allows you to specify a folder on your company (typically already containing relevant files) that the agent can work inside. The agent has access to all files in the folder and can create files in there. E.g. Claude cowork, check the video. https://www.anthropic.com/product/claude-cowork

OpenClaw is more aggressive. You can give the agent full access to your computer, e.g. all files, browser, email, message apps etc. This wide range of access is a key factor that made OpenClaw popular, but as you can imagine, this can also break things.

3. Product-specific AI

E.g. Notion AI, Zoom AI Companion, Adobe Acrobat AI Assistant, Gemini in Google Doc These products are tightly coupled with a specific existing product. The goal is to make using this existing product easier. E.g. Notion AI can help you summarize/clean up/create Notion pages. Typically these tools are not general: they can only be used within the product.

4. Coding agents

E.g. Claude Code, Cursor, GitHub Copilot (not to be confused with Copilot. See later section) Products that specialize in software engineering tasks.

Similar to Computer / local-work agents, these agents have direct access to your computer. This makes writing code easier than using general chat assistant. You can use the latter to write code, but you would need to manually copy and paste code from the chat window to your local code files. Coding agents such as Claude Code has direct access to your code files, which makes e.g. editing multiple code files easy.

Note: if you are not looking to do software development, these tools are in general not relevant to you.

5. Agent builders

E.g. N8n, Zapier, Langchain, CrewAI

This and the next category of products are tools to build agents. So far we have been looking at agent products that were already built. When you use ChatGPT, Claude, you are using an AI product where someone else already chose the LLM model, tools, and user experience.

This category of products allow people to build their own agents (e.g. we can build our own ChatGPT-like app). The advantage of building your own agent is customization. E.g. your company uses a specific document system thats not supported by ChatGPT, so you might want to build a custom agent to connect with this system. In general, if existing AI products cannot support your use case, consider building your custom agents.

Within agent builders, some are no-code, which does not require writing code to build agents, such as n8n, Zapier. Main interface is drag-and-drop style.

Some are “agent frameworks”, e.g. Langchain, CrewAI, which require writing code to define custom agents. Generally more flexible/powerful than no-code solutions, but require coding.

Where things get messy

Copilot: this is probably the most confusing term of them all (thanks Microsoft). It's not one product, but a naming pattern used across Microsoft/Github products (take a deep breath):

• Microsoft Copilot - consumer/general AI assistant
• Microsoft 365 Copilot - AI for Microsoft work apps like Word, Excel, PowerPoint, Outlook, Teams, and Microsoft 365 work context
• Copilot in Word / Excel / PowerPoint / Outlook / Teams - product-specific AI inside Microsoft apps
• Copilot Studio - tool for building/configuring agents
• GitHub Copilot - coding agent

I don't know about you, but Im confused (I wonder if Microsoft people are confused).

OpenAI Codex (desktop app): OpenAI’s mega app

• Started as a coding agent (similar to Claude Code). Recently added support for Computer / local-work agents: ability to work in a folder in your computers / use your browser.

In general, AI apps kpt expanding. E.g. claude adding support as a plugin in Excel app. The trend is that everyone is trying to make an app that does everything.

Cheatsheet