Five AI Experiments You Can Run Today

The December issue of this newsletter about AI capabilities that changed in 2025 was the most-read of the year.

But here's the thing: reading about AI capabilities and actually using them are completely different.

So I spent the last week building something to help you further, a lab where I will help you try each capability yourself. Five experiments. Thirty minutes each. Copy-paste prompts included.

No course to buy. No theory to absorb. No vague promises about "AI transformation."

The gap between reading about AI capabilities and actually testing them is where most people get stuck. Companies spend months debating which tools to adopt when they could just run a 30-minute experiment and know for sure.

This experimentation ‘lab’ I created closes that gap. Five experiments. Thirty minutes each. Very quickly, you'll know which capabilities work for your workflow.

The Five Experiments

Each one covers a capability that changed in 2025, not theory, but tools you can use right now. NotebookLM went from "interesting research tool" to "instant deck generator." Claude Skills turned prompts into reusable workflows. ChatGPT Projects made team context persistent. Nano Banana finally (!!) created legible text in AI images. Perplexity Spaces became living research hubs.

Here's what you can try:

Slide Decks + Infographics
Turn any document into presentation-ready slides in 30 seconds using NotebookLM. I tested this with a 20-page strategy doc. I got back a 5-slide executive version and a 12-slide team version. No manual screenshot cutting. No reformatting. Just upload, generate, done.

Reusable Workflows
Encode your recurring tasks as Claude Skills so they run consistently across your team. This solves the "good prompts in Notion that nobody uses" problem. Instead of documenting processes, you build workflows people can actually repeat without thinking about it.

Team Workspaces
Set up persistent context in ChatGPT Projects where your team can work asynchronously without losing the thread. Replace those email chains where the 5th person asks the same question that was answered in message 2.

Production Visuals
Generate diagrams, mockups, and campaign assets with Nano Banana Pro. The first AI image model that creates clean, legible text. Actual design-ready outputs you can use.

Living Research
Build research hubs in Perplexity Spaces that accumulate insights over time instead of starting from scratch with every question. Your competitive intel stays current automatically instead of going stale in a deck.

What Makes This Different

Each experiment has the same structure:

• What the capability is and how you can use it

• An example prompt to use (copy-paste ready, no modifications needed)

• What you'll get (specific output)

• Success criteria (you'll know if it worked or needs iteration)

There's also a "common mistake" callout for each one - the thing that trips people up when they first try it. I learned these by watching teams test these capabilities. Saves you 20 minutes of troubleshooting.

How to Use the Lab

Pick the capability that solves your biggest current friction point, the one that would free up the most time or remove the most manual work.

Try that experiment this week.

If it works, you've freed up capacity. If it doesn't, you've learned what "production-ready AI" looks like in your context. Either way, you're ahead of where you were Monday.

Then, if it works, figure out how to make it operational. Can your team use this consistently? Where does it fit in your existing workflows?

The lab shows you what's possible. That next step, implementation across your company, is what comes next.

Try the experiments: lab.atomictheory.ai

Need help making this operational?

These experiments show what's possible. Implementing them systematically across your company, figuring out which capabilities map to your processes, running successful pilots, building the foundation to scale what works - that's where I can help.

You can work with me directly through Atomic Theory for AI guidance and implementation, or talk through your specific challenges with Atom, my AI assistant who knows my frameworks and can help you get started, before you decide to speak with me.

Learn more: atomictheory.ai

Or just reply to this email. We can talk through what makes sense for your situation.

The Atomic Builder is written by Faisal Shariff and powered by Atomic Theory Consulting Ltd — helping organisations put AI transformation into practice.