There's a new AI model out this week. If you spend any time on LinkedIn or in a founder group chat, you already know. Anthropic launched Fable 5, and the timeline did what the timeline does. Charts. Screenshots. A dozen people telling you everything is different now.

And here you are, running a real store, with a Klaviyo account that needs attention and a Meta account that isn't doing what it used to, wondering if you're supposed to drop everything and go learn this.

Short answer. No.

Longer answer. There's a small, specific part of this that's genuinely useful to you, and a much larger part that's noise. Let me sort the two.

Most of what you're reading doesn't apply to you

Here's the thing about a model launch. The headlines are written for engineers and investors, not store owners. You'll see benchmark charts where the new model beats the old one by a few points. You'll see it write code faster, read scientific papers, do clever things in a lab. Impressive. Genuinely.

None of it tells you whether your email is pulling its weight, or where your margin is going.

There are even bits walled off completely. This one blocks anything near cybersecurity or biology and hands those questions to a smaller model. Doesn't matter to you. And it's priced right at the top of the market, so it isn't the thing you casually leave running on everything. Worth knowing before you build half your workflow on it.

So park the hype. The real question is narrower: what can a tool like this actually do for a brand like yours, today?

The useful part is narrow, and it's real

Here's where these models have quietly got good. Not at replacing you. At reading the mess you don't have time to read.

Your business throws off data all day. Klaviyo, Shopify, Meta, your P&L, your supplier invoices. Most of it sits there unread, because reading it properly takes hours you don't have. A capable model is very good at being the person who reads it and tells you what stands out.

That's the shift. It used to take a sharp analyst to look at your numbers and say 'this is the bit that's off.' Now you can get a decent first pass in minutes. Not a final answer. A first pass. There's a difference, and I'll come back to it.

Six things you can actually do with it this week

Pick one. Don't try all six.

Read your Klaviyo and find the gaps. Export your flow performance, paste it in, and ask: 'Here are my Klaviyo flows and their revenue. Which flows am I missing, and which existing ones look underbuilt for a brand my size?' You'll get a punch list in a couple of minutes.

Separate a profit problem from a sales problem. These are not the same thing, and founders blur them constantly. Paste your P&L and ask it to walk the numbers in order, from revenue down to what you actually keep, and show where the money leaks on the way. Flat sales and thin margin need different fixes. Work out which one you've got first.

Get to your true CAC. Give it your ad spend, your new customer count, your average order value, and your repeat rate, and ask for your real cost to acquire a customer and your LTV to CAC ratio. Most founders are running off a number that's politely wrong. This gets you closer.

Make sense of a dashboard. Screenshot your Triple Whale, Meta, or Shopify dashboard, drop it in, and ask what changed and what's worth worrying about. The newer models read images well now, so this actually works.

Draft the flow you keep putting off. The winback. The post purchase sequence. Ask it to write a first version based on your products and your customer. You're not shipping it blind. You're reacting to a draft instead of staring at a blank screen, which is far easier.

Write the thing you keep avoiding. The SOP for the task only you know how to do. The job description for the role you need but can't quite define. Talk it through to the model and let it structure it. That's an afternoon of avoidance gone.

The model was never your bottleneck

Now the part everyone skips.

You can have the best model in the world open in a tab. It does nothing for you until you know what to ask it. That's the actual skill, and it has nothing to do with which model launched this week.

Go back through that list. Every one of those prompts works because it points at a real problem you already have. The founders getting value out of these tools aren't the ones with the newest model. They're the ones who got clear on the one thing that's genuinely stuck in their business, then pointed the tool at that.

And look, that's the bit no model gives you. Knowing whether your problem is acquisition, conversion, average order value, retention, or margin, and being honest about which one. The tool is fast. It isn't wise. The wisdom is still your job, or ours.

So here's what I'd do this week. Don't go signing up for the newest thing and feeling productive for an afternoon. Pick the one number that's been stuck the longest. Then use whatever AI you already have open, you don't need the brand new one, and point it straight at that number. Start there.

That's the whole game. Less chasing tools, more pointing them at the right problem.