🗞️ RUFUS is now Mainstream 💥💪

Welcome to the 10th edition of the AI for E-Commerce newsletter, your trusted resource for actionable AI strategies in eCommerce, especially on Amazon! We’ve hit 1200 subscribers (up a 100 from last week)💥🔥!

So, each issue has 4 sections:

1. Practical AI Use Cases

2. Cool tools that you can start using today to improve your eCommerce business

3. Nerd Bytes for all you nerds out there

4. Voices from AI Thought Leaders in the E-Commerce space

Our focus is on tangible benefits. No technical jargon, no pie-in-the-sky concepts—just straightforward, actionable advice to help you stay competitive and grow your online business. 

Let’s go!

For this first section on AI Practical Use Cases, we are talking about RUFUS entry on to Amazon’s desktop version! Your new shopping assistant is now just a button away.

And, has also made its way into the search bar!

The last three searches in the autocompletes options are actually prompts that open up RUFUS in a chat window in the bottom left of the search results page.

Interacting with RUFUS feels similar to engaging with another large language model, but it is concise and always directs you to either:

• Amazon product suggestions, or

• Amazon searches on specific keywords

For example, this prompt provided a brief paragraph of information followed by a clickable search for “Baking Desserts with Unsweetened”

If I examine the URL of the resulting search page it already has those keywords pre-filled:

https://www.amazon.com/s?k=Baking+Desserts+With+Unsweetened+Chocolate&niid=nl_cl_lst_c_1_0&nrid=HT27J3Z62RBY0WN6PTWN&qis=NileBrowseNode&ref=nl_cl_lst_c_1_0

At the end of each interaction, RUFUS consistently presents more pre-curated prompts for you to explore further.

RUFUS appears to have unlimited suggestions to keep the conversation engaging and dynamic.

Delving deeper into these "prompt-holes" may lead to increasingly specific inquiries, typically framed as 'how to' questions.

Let’s look at the third search of these choices “Cooking Savory Dishes”. It leads to this URL:

https://www.amazon.com/s?k=Cooking+Savory+Dishes&niid=nl_cl_lst_c_1_4&nrid=9F5SAF6M0ZJ8CZNQ8DAB&qis=NileBrowseNode&ref=nl_cl_lst_c_1_4

Notice that this page has books, not actual products.

I repeated my search using the original prompt: “What are common unsweetened chocolate chip uses” and received a completely different response. This demonstrates that RUFUS is capable of generating non-monotonous, creative alternatives.

This time, RUFUS provided a detailed response, concluding with a list of clickable options. These links lead to new keyword searches for:

• Unsweetened Chocolate Chips

• Baking Chocolate Bars

• Unsweetened Cocoa Powder

• Semi-sweet Chocolate Chips

So, what are the implications of having RUFUS influence the shopper journey?

Here’s what I learned: The paths shoppers can take with RUFUS are infinite, each uniquely shaped by their preferences and choices.

So how do we prepare for so many different prompt paths?

• Monitor Auto-Generated Prompts: Start by paying close attention to the types of questions RUFUS auto-generates. Understanding the common themes or topics can help you anticipate shopper interests and tailor your listings and infographics to better intercept buyer inquiries.

• Explore Prompt-Holes: Actively explore different prompt paths to discover the types of products or content that shoppers are being directed towards. This can provide insights into shopper behavior and preferences.

• Analyze Search Query Performance: Regularly review your Search Query Performance data to monitor if there are any obvious drops in performance. This will help you identify if you are losing visibility on your most important keywords and need to adjust your SEO strategies or listing.

So far, RUFUS has not showcased any sponsored listings, only organic ones. But in the future this might change. Keep an eye out!

For the Cool Tools section today, we are talking about a cool new Canva AI tool for replacing the background of any image with an AI generated one. It’s called Replace Background and can be located under the Apps section.

To use this tool upload any image and swap out the background using an AI prompt.

I uploaded the image on the left and wrote this prompt “Lush green Georgian countryside”. It automatically figured out that the cans needed to be placed on a somewhat flat surface, and also figured out the slant of the shadow.

To use this tool, you will need a Canva Pro account. I pay $30 a month for 3 seats on Canva Teams. Honestly, with over 200 AI add-ons available, it's definitely worth it!

In this week's Nerd Bytes section, I want to highlight a couple of cool updates to ChatGPT I've noticed.

First up, when it digs into topics that might need some real-time web browsing, ChatGPT now throws in clickable URLs for its references. Super handy!

And here's another cool feature I spotted: now you can highlight any part of ChatGPT's response, and a "Reply" button pops up. Hit that, and you can dive deeper with follow-up questions right on that specific section.

I wouldn’t be surprised if OpenAI got the idea from social media. This really smooths out the conversation!

For the Thought leader section this week we asked our friend Anthony Lee from A.I. Plus Automation about the future of AI in E-Commerce and this is what he had to say:

“With AI at the top of everyone's mind, it is no wonder the Amazon world is abuzz about Rufus, the conversational AI shopping bot.

Rufus is not only innovative, but for clever sellers its use will lead to valuable insights into how it will help shoppers, ideally, find their products.

The thing is, Rufus is a chatbot, built to make the shopping experience more enjoyable for customers. That doesn't necessarily have anything to do with how an AI may choose to help the algorithm rank and sort search results.

While Amazon is, as always, tight-lipped about how the A9 algorithm works, the closest name we can attribute to search-algorithm-related language models is BERT.

Multiple science papers have been published about BERT-assisted search result engines (like AttriBERT, which utilizes AI to generate behavioral product attribute preferences in session). Whether or not these BERT specific tools are being deployed by Amazon, it remains clear that the future of the Amazon search algorithm is AI assisted.

But what does that mean, exactly?

It means that things like keywords, key phrases, stemming, plurals, misspellings and all the other iterations of search terms that impact varying browse pages will no longer be as important. With AI assisted search, intent and semantic relationships are the key factors.

Ok, so what does that mean for sellers?

The future AI assisted search engine is then going to be responsible for indexing and ranking listings based on their own understanding of what is in Amazon's catalog, so if you want your listings to have the greatest chance at being visible and unambiguous to the AI's indexers, they need to be readable by AI.

One way to work toward this is to use Amazon's own AI tools.

For example, to give your listing images the best shot at being "seen" by the future AI search engine you could use the Amazon AWS Rekognition tool. This tool looks at images and tags the visible objects within them, then assigns a probability score to how accurate the tags are likely to be.

This exercise would help you choose only the clearest and most obvious (to AI) images for your listings (by choosing only those with accurate tags).

Another tool you can use is AWS Comprehend. This is Amazon's AI tool that analyzes text and identifies "entities" (much like Rekognition with tags). When you input your listing text, Comprehend will tell you all the "entities" or otherwise the obvious keywords it recognizes. This exercise would allow you to use only the most optimal wording and word order to ensure AI clearly recognizes the keywords in your listing copy.

All this leads to your products being easier to find and fetch by an AI search engine.

And the best part? These tools are nearly free. All you need is an AWS account.“