Google's DreamBeans Is Fun. Here's Why I Ignored It and Built Something That Actually Moves the Needle.
Google just dropped a tool called DreamBeans that turns your photos into cartoon versions of your life. Genuinely charming. My kids would love it. I spent about four minutes with it before closing the tab and getting back to work.
That's not a criticism of Google. It's an observation about where most small business owners are spending their AI curiosity right now — and where they're not.
Every week there's a new tool with a quirky name and a satisfying demo. Every week a handful of business owners try it, post about it, and move on. And every week the gap widens between people who are using AI to build something durable and people who are using it to make content about using AI.
I want to talk about the other side of that gap.
The "Shiny Tool" Trap Is Real, and It's Costing You Time
There's nothing wrong with exploring new AI tools. Curiosity is a legitimate part of staying sharp. But curiosity without a filter is just procrastination with better branding.
DreamBeans is a perfect example of a tool that's easy to demo and impossible to justify in a business context. It's consumer entertainment dressed up as innovation. The TechCrunch headline calls it Google's "weirdest-named AI tool to date," which tells you everything about the intended audience.
The question I ask before touching any new AI tool is simple: does this save me time on something I already have to do, or does it create a new task I didn't need? DreamBeans creates a new task. So it stays in the tab graveyard with the other interesting-but-irrelevant experiments.
Your time is the constraint. Not your access to tools.
What I'm Actually Building Right Now
While DreamBeans was making headlines, I was deep in a workflow I've been refining for the past few weeks: a client onboarding sequence built inside n8n that pulls from a Notion intake form, generates a customised welcome pack via Claude, and delivers it via email without me touching a single button.
This took real time to build. It wasn't satisfying in the way a cartoon selfie is satisfying. But it means every new client I bring on gets a professional, personalised first impression within minutes of signing, regardless of where I am or what I'm doing.
That's the kind of build that compounds. A cartoon version of my face does not compound.
If you're a coach or consultant, think about what your current onboarding looks like. Is it manual? Inconsistent? Something you dread when you're already busy? That's the exact right target for an AI build. Not because it's glamorous, but because it's costing you time and energy every single cycle.
The Framework I Use to Decide What's Worth Building
I've got a simple filter I run every time someone sends me a tool to try or I stumble across something new. Three questions, in order.
First: is there a real task this replaces or speeds up? Not "could I imagine using this" but "do I currently spend time on this specific thing?" If the answer is no, the tool goes on the maybe list and I move on.
Second: how often does that task repeat? A one-off task that takes an hour is a poor candidate for automation. A task that happens weekly or with every new client is a strong candidate. Frequency is where automation earns its keep.
Third: what breaks if I get it wrong? Some tasks are high-stakes. A mis-generated contract or a wrong email to the wrong client is a problem worth taking seriously. Some tasks are low-stakes. I build for low-stakes first, get confident with the tooling, then move into higher-stakes workflows carefully.
DreamBeans doesn't survive question one. Most trending AI tools don't. That's fine. The filter is doing its job.
The Actual Opportunity Right Now
Here's what I think is true about this moment in AI: the tools that matter for small business owners are already available. Claude, ChatGPT, n8n, Notion, Make. The gap isn't access to tools, it's knowing what to build with them and having the discipline to build it instead of chasing the next demo.
The business owners I see pulling ahead aren't the ones who've tried the most tools. They're the ones who picked three and built something real with them. A workflow that saves four hours a week. A follow-up sequence that runs automatically. A client intake process that doesn't depend on them being in the office.
That's what "AI for business" actually looks like. Not a cartoon. A system.
Your Move
Pick one task you do manually that happens at least once a week. Write it down. Then open Claude or ChatGPT and describe it in plain language: "I do this task, it takes this long, here are the inputs and the outputs." Ask it to help you map a simple automation for it.
You don't have to build it today. But writing it down is the first step toward it running without you. That's worth more than any tool with a fun name.