Announcing o11's backing from Y Combinator.
Agents vs. Chatbots: Why 'Doing' is Better than 'Chatting'
We are witnessing a fundamental shift in the AI landscape: the transition from “Chatbots” to “Agents.”
For the past two years, the world has been fascinated by Chatbots (LLMs). They are brilliant conversationalists. They can write poems, explain quantum physics, and debug code. But they are passive. They live in a box, and their only output is text.
Agents are different. Agents can use tools.
The Functional Distinction
| Capability | Chatbot (e.g., ChatGPT) | Agent (e.g., o11) |
|---|---|---|
| Input | Text Prompt | Text Prompt + Environment State |
| Action | Generates Text | Executes Commands |
| Output | “Here is a recipe.” | “I have ordered the groceries.” |
| Workflow | User copies/pastes result | System updates the file |
Why o11 is an Agent
o11 is built on an agentic architecture. When you ask it to “Make this slide blue,” it doesn’t just tell you how to change the background color. It:
- Observes: Inspects the current PowerPoint selection.
- Plans: determining the correct API call to modify the fill property.
- Acts: Executes the code to change the color.
- Verifies: Checks that the change occurred.
The Agentic Future of Work
In an enterprise setting, you don’t need a conversation partner; you need a force multiplier. You need an entity that can take a high-level goal (“Build a model”) and execute the thousands of low-level steps required to achieve it.
Stop chatting. Start doing.
