Stop Asking "Which AI?" — Start Asking "What Tools Did You Give It?"

The question I get most at conferences, in my classroom at VCU Brandcenter, and from clients is some version of "What AI tool do you use?" I used to answer it. ChatGPT for this, Claude for that, Gemini when I need the other thing. But I've stopped, because I think it's the wrong question.

The better question is: what tools did you give your AI?

Most people interact with AI through a chat window. You type a prompt. You get text back. Maybe you copy it into a doc, tweak it, move on. That's useful, but it's a fraction of what's possible now. The bigger change happening in AI isn't which model you pick. It's what you connect that model to.

An AI that can access your file system, open a browser, create a slide deck, pull live data, and write directly into your productivity apps is a completely different animal than the same AI in a chat window. Same engine. Wildly different capability.

Asking "Do you use ChatGPT or Claude?" is like asking a carpenter "DeWalt or Makita?" Not a useless question. But it misses the point. The more interesting question is what's in the whole workshop, and whether the carpenter knows how to set it up.

Same AI, different tools, outcomes

That's what I've been exploring in my own work. Instead of writing better prompts, I've been giving AI access to tools: file systems for reading and creating documents, browser access for live research, connections to productivity apps, and structured instruction files that teach it how to do specific tasks. The difference between "describe how to make a presentation" and "make the presentation" isn't a better prompt. It's giving the AI access to the software that builds presentations.

This matters for how we teach and talk about AI in media. Most AI literacy conversations, in classrooms, newsrooms, agencies, still center on prompt engineering and model selection. That's important, but incomplete. It's like teaching someone to type and calling it computer science. The next competency is "equip your AI with the right tools for the job." That's a design problem, not a language problem.

I'll be honest, this is still early. The tooling changes fast (OpenAI and Google are just now playing catch up to Claude Cowork), the setup isn't always intuitive, and best practices are still forming. But the value of AI is less and less about the model itself and more about what you let it access and do.

So next time someone asks what AI you use, try flipping the question: "What tools did you give it?"

By: Jeff MacDonald

Affiliate Fellow, Media + AI

Note: This post was developed with the assistance of Claude (Anthropic, 2026), used as a writing collaborator. More on my agentic process here.