I'd go back to Amazon. With one condition.

I loved working at Amazon.
The culture, built on the Leadership Principles, was incredible. Focus on results for the customer. High bar. Wicked smart people. Narratives instead of slides. I still use those habits every week.
Listen to this post on Substack.
But parts of it taxed me.
If I went back today, I think I’d enjoy it a lot more. Because I’d put AI on exactly those taxing parts.
Here’s the top of the list.
1) Writing PRFAQs.
PRFAQ stands for Press Release / Frequently Asked Questions. If you’ve worked at Amazon, you’ve probably lived this. You bring up an idea, and your teammate or your manager says, “That’s a great idea. How about you write a doc?”
Yes. Write the press release for a product that doesn’t exist yet. Then answer every question a skeptical VP could ask. Then rewrite it. Then rewrite it again.
Now there are skills for this. Hand Claude the prospective customer, the problem and the data, and you get a first draft that already follows the format. You spend your energy on the thinking (the big idea), not the scaffolding.
2) Writing “crisp.”
If you’ve been there, you’ve heard it: “Can you make this crisper?”
No weasel words. Numbers, not adjectives. Every sentence earns its spot.
Great standard. Exhausting to hit at 2 AM.
Also a skill now. Fun fact: this post went through one before you read it. 👀
If I could go back, I’d collect everything written by the person who kept telling me “make it crisper” (let’s call them Bob) and build a bob-crisper skill. Before I handed Bob anything, I’d run it through the skill so it already cleared Bob’s bar. Or at least made it more likely.
3) Writing bridges.
Forecast said X. Actual came in at Y. Explain the gap to within ±5% before finance skins you alive.

Pull the data. Slice it six ways. Find the two or three drivers that actually matter. Compare it to every other bridge ever shared because finance WILL REMEMBER and cross examine you. Write it up ‘crisply’ so a leader gets it in 30 seconds or less.

With all the fires, I never had time to build something to automate it. And honestly? It takes so much judgment I’m not sure I could have.
Which is exactly why it’s the perfect AI use case.
AI is awesome at the grind: pull, slice, rank the drivers. It’s good enough at first-draft judgment that you’re reviewing a bridge instead of starting from a blank page. Especially when you’ve given it good context, like skills. You still own the call. You just stop paying for it with midnight oil.
A note for leaders.
That cross-examination from finance was miserable.
(Sorry. I meant “Diligence.”)
It was also the single best driver of forecast accuracy I’ve ever seen. Nobody wants to explain the same miss twice. So people forecast more carefully the next time.
If you review forecasts and you’re unhappy with accuracy, don’t reach for better tech first. Put on your bad cop hat. Ask “why?” until the answer is real. Remember last quarter’s bridge.
The culture didn’t need to change for me to enjoy Amazon more.
Just the tooling did.
Well... maybe.
Next up: some problems I faced at Amazon that AI can’t solve. (Spoiler alert: promotions.)
Let’s Turn Your Data into Growth
No more guesswork or missed opportunities.


