2024
AI agents go mainstream
AI agents — LLMs that independently use tools, make plans, and execute multiple steps — go mainstream. Claude Computer Use shows AI operating a computer.
From answering questions to taking action
2024 was the year AI moved from generating text to taking action. AI agents — systems that use a language model as a reasoning core but can autonomously call tools, browse the web, write and execute code, manage files, and chain together multiple steps to complete complex tasks — moved from research demos to production systems used by millions.
Claude Computer Use
In October 2024, Anthropic launched Computer Use in public beta — a capability allowing Claude to directly control a computer: moving the mouse, clicking, typing, opening applications, and browsing the web. This was qualitatively different from chatting or code generation: the model could now autonomously navigate graphical user interfaces designed for humans. Anthropic positioned it as a tool for automating tedious tasks but also acknowledged the safety challenges of an AI that can take actions in the real world.
The agent ecosystem
Throughout 2024, an ecosystem of agent frameworks exploded: LangChain, AutoGPT, CrewAI, Microsoft AutoGen, and dozens of others. OpenAI launched Assistants API with code execution, file access, and function calling. Enterprises began deploying agents for customer service, code review, data analysis, and document processing. The shift from AI as a tool you talk to, to AI as an agent that acts on your behalf, was well underway.