Generative or agentic?

Generative: it writes

• One request, one answer. You ask, it produces text or an image, and it stops. Nothing happens until you type again.
• You are the loop. You read the draft, decide what is wrong and tell it what to fix.
• Everyday use: every plain chat in Gemini, ChatGPT and Grok. This is most of what people need.
• The risk is a confident wrong sentence. The defense is the source attached and your read.

Agentic: it acts

• A goal, then many steps. It plans, searches, reads what it found, decides the next step and repeats until done.
• It uses tools: a policy search, a file writer, a schedule. The model is the same; the tools and the loop are new.
• In practice: Deep Research, Gemini agents with connectors, Grok skills that build a file, automations on a timer.
• The risk is a wrong step you never saw, or an instruction hidden in a document it read.

What changes for you

1. Give a goal and a stopping point. “Find the governing instruction, cite the paragraphs, stop and show me.”
2. Read the steps, not only the result. Deep Research and Grok show what they searched and opened.
3. Expect minutes, not seconds. A file from a Grok skill can take a quarter of an hour.
4. Nothing leaves without you. No agent here sends email or signs. You still release the product.