Turns out, LLMs are super useful. Your level of knowledge on their architecture and makeup is the limiting factor.

I’m not an AI expert. Heck, I’ve barely met the guy! 🤣 Legit though, I’m not an engineer by trade and I don’t know how to code. For the sake of full transparency, I can barely write SQL. I consider myself solidly in the marketing and advertising space. 

But I’m what you call a figure-outerer (very technical term). God made me to be an Adventurous Maker. What does that mean? I like to build things and live an exciting life while doing it. Exciting doesn’t always mean successful. Lots of challenges and what look like failures. Let’s just call those opportunities for now. And yet I keep getting my hands dirty in order to figure stuff out.

6 months ago, using an LLM was a weekly thing for me. 12 months ago, monthly. Today, it’s multiple hours a day either reading outputs, cleaning up current systems, like scheduled tasks, or building new skills.

There are a couple levels of LLM users:

1. The coworkers: they open up a chat, ask it questions, and have a conversation. They get some answers faster than they normally would and that's helpful, more often than not.

2. The department heads: they've got a deeper understanding, they're creating skills, scheduled tasks, building artifacts, spinning up reports faster, and discovering it's capabilities more and more each day.

3. The executives: these people are truly creating with the LLM, building self-learning agents, using agents to control other agents and run skills, and doing seriously heavy lifting of their workload.

The main difference between all three is who's driving. The coworkers treat the LLM like an Uber, whereas the executives are driving the LLM, even supercharging it over time. They see it as a vehicle that gets them from point A to point B, not merely something to experience, i.e. the modern day taxi ride. They're aggressive in their usage. That's next level.

In the same way that most people aren't going to just go start their own company in order to have a job, most people aren't going to spend the time trying to understanding an LLM, studying how other people use it, taking courses, or just risk wasted time being confused attempting to build.

That takes a unique kind of person, a visionary, just like it takes a unique kind of person to follow the directions of a visionary to make the work actually happen. I'm not here to debate which is more important.

My point: people driving the LLM are paving the way for the LLM coworker users, and we simply don't need as many of the executive level users. The coworker users, however, are going to be using it the most on a daily level as part of their workflow.

No, I don't believe that AI is going to take over the world. Sorry, Elon. We've got a long way to go before iRobot becomes true.

LLMs as really incredible tools. They're going to help make us better, more efficient humans. There's a naivety in that, certainly, but it's my newsletter. 

What does this have to do with Amazon?

I'm so glad you asked.

Right now, SaaS companies are connecting to LLMs that are allowing you to query their large datasets, making your life easier. That looks like "show me all the search terms that haven't converted in the last 30 days," or "help me collect market share data for these 5 brands."

This is really only the beginning. We were previously beholden to SaaS companies' features and viewpoints. We're headed to a place where you use SaaS companies with LLMs based on your parameters that get you to your goal. You'll be in charge.

The best way to get started...

Talk with your current service providers and see if they have an MCP connection, a plugin, or any other way to connect their software and data to your current LLM. Then find the intersection in what they're building and what you're currently doing.

That's it. It doesn't take you getting a new degree, teaching yourself to code, or even attending a webinar. Build with what you've got right in front of you.

So, what are you waiting for? Go start being a figure-outerer too.Could you imagine if this were a real race? It would be amazing.

Amazon’s business operations falls into three distinct categories: Catalog, Creative, & Advertising. I create a content tasting flight of smaller yet no less delicious samplings of that Amazon ecosystem. Today’s Flight focuses on the basics.

Daily scans of your Catalog by a faster more capable machine than yourself means less time wasted scanning everything before you get to the real priority.

Use Helm. Have I really not been obvious enough about this? Try it now. Seriously. Do it.

Ever frustrated by how long Amazon A/B Experiments can take? If only there was a way to test your test before you tested. Well, there is.

Get both versions of your creative and run a test on PickFu before you run the A/B Experiment. This tried and true software allows you to poll Prime Members, or specific demographics, to see how they’ll react to creative, among other things. You can also get more qualitative data as well, so you've got a why not just a what.

The Subscribe & Save Dashboard got a bit of an upgrade. CLTV by Segment now shows One Time Customers vs. Reorder Customers vs. Subscribers. The Total Sales & Subscribe & Save Sales by Number of Deliveries sections are both great indicators of whether or not you have a retention problem.

Review the data and go create a custom AMC audience that turns the reorder customers into Subscribe & Save customers. For any renewable business, the best way to grow your business long term is to find ways to turn one time buyers into repeat buyers. First time Subscribe & Save order coupon anyone?

Espresso Martinis were super hot, and then not. The hip cocktail crowd loves to shit on what people like. It’s not for me to tell you what to drink or how to enjoy it. If you like your bourbon neat, great. If you want it over 1,000 ice cubes, fine. I wouldn’t drink it on top of a glacier but drink it how you like.

This Espresso Martini riff came from my diving into the Sidecar and it’s variations in Death & Co’s Cocktail Codex book, which inevitably had me using Cognac more. I also had some ingredients I wanted to use up and voila, another kitchen sink drink miracle.

Every two weeks I release a new podcast episode that talks all about the future of retail. This is the most recent one.

Amazon is complex. And the ability to synthesize and then analyze data takes a lot of time. GenAI has allowed Ritu to cut that time down significantly and focus more on creating solutions.