- Published on
An opinionated guide to which AI to use to do stuff
- Authors
- Name
- DeepHorus
A new opinionated guide on selecting AI tools highlights a major shift from simple chatbots to agentic systems capable of performing hours of human work autonomously. Updated as of July 27, 2026, the guide by Ethan Mollick reflects the rapid evolution of AI capabilities over the past year, moving beyond conversational models to sophisticated agentic workflows.
The guide traces how the AI landscape has transformed. A year ago, recommendations centered on chat-based models like ChatGPT, Claude, and Gemini, with specific versions such as o3, Claude 4 Opus, and Gemini 2.5 Pro. Today, the focus has shifted to agentic systems where AI can execute complex tasks independently. Notably, Gemini has been dropped from Mollick's list because Google still lacks an established entry in the Codex/ChatGPT Work/Cowork category, and its Gemini Spark model has yet to prove its value.
Key details from the guide explain how users can give AI access to a computer. In ChatGPT, the mode is called Work; in Claude, it is Cowork. These allow AI to operate within the companies' provided computers. For deeper integration, ChatGPT offers Codex and Claude offers Code modes, which let AI access the user's own computer. The naming is confusing: Work and Cowork on mobile differ significantly from the same modes on desktop apps, where they essentially become streamlined versions of Codex.
The implications of this shift are profound. Agentic AI can now perform the equivalent of many hours of human work in a single session. However, the unintuitive naming and differing capabilities across platforms may hinder adoption. The guide emphasizes that the most powerful use is giving AI access to your computer, but users must navigate a complex web of modes and features.
Reactions to the guide note its value in cutting through the confusion. Mollick’s perspective helps users understand which AI tools are best suited for specific tasks, acknowledging the trade-offs between simplicity and power. The blog post also links to recent articles on OpenAI's accidental cyberattack and a fireside chat with the Claude Code team, suggesting ongoing developments in the field.
Looking ahead, the trend toward agentic AI is expected to continue. As models like ChatGPT and Claude evolve, users will likely see more seamless integration of autonomous capabilities. However, improving clarity in product naming and functionality will be crucial for broader adoption. The guide serves as a useful snapshot of a rapidly changing industry, where staying informed is key to leveraging AI effectively.
