We’ve talked a lot about the downsides of shopping online. You can’t touch or feel the item. You can’t try it on. You can’t really know what it looks like in person.
But over the past few years, there are more and more things we can do online. Now, you can try on clothes virtually. You can know exactly what material it's made from, its style, and whether your friend might like it.
And soon, you might even be able to touch it virtually, even if it's not in front of you.
In the meantime, we’ve got the next best thing: having a personal shopping assistant in store.
Amazon Rufus launched in the US in early 2024 and has gradually rolled out to other countries.
What is Amazon Rufus?
Amazon Rufus is your AI-powered shopping assistant on Amazon. It’s named after Amazon’s first two employees’ corgi—one of the cutest dog breeds in the world.
Rufus is built using a custom Large Language Model (LLM) trained on Amazon’s extensive product catalog, customer reviews, community Q&A, and web data.
What can you do with Amazon Rufus?
Rufus is designed to help customers find and decide on products more easily while shopping on Amazon.
It can’t solve math problems or write you an essay. But it can give you all the information you need to make confident purchases.
I won’t go into detail, since you can read it on Amazon’s article.
But essentially, you can:
- Learn what to look for when shopping for specific items. Rufus provides insights into key features, factors to consider, and tips for choosing the right product based on your requirements.
- Compare products to understand which one suits you better, with side by side comparisons. Rufus can tell you that while merino is less warm than cashmere, it’s more breathable and durable, making it more suitable for hiking (if you hike).
- Get product recommendations tailored to your queries. Rufus analyzes customer reviews, popularity, and product attributes to recommend items based on your query.
- Ask detailed questions about specific products. You can ask it if a hand grinder can prepare coffee grounds fine enough for espresso, or if a dog bed can fit a medium-sized corgi.
- Check historical prices to understand if an item on sale is truly cheaper than it was.
What we found about Amazon Rufus
All the possibilities around Amazon Rufus sound great. And if executed well, this feature can really improve the customer experience, boost product discoverability, and bring in more revenue for Amazon sellers.
Amazon also projects that Rufus would indirectly contribute over $700 million in operating profits in 2025.
It got us really excited. So we went ahead to ask Rufus some questions to see how it fares. Here’s what we found.
Rufus recommends but doesn’t explain
When we asked Rufus “What’s a good carry-on backpack that fits under airplane seats?”, we were expecting it to share what are factors to consider.
But it didn’t.
It went straight into recommending a few backpacks without telling us why.
Rufus lists, but doesn’t really compare
We got Rufus to compare the Bose QuietComfort Ultra and Sony WH-1000XM5. It provided a paragraph listing the features of each, without actually comparing them side by side.
When we asked follow-up questions like “Which headphones are better overall”, it gave us 2 other headphones, completely different from the 2 we asked it to compare. (we asked the same question and ChatGPT knew we’re asking about the same 2 headphones.)
Rufus has your purchase history but doesn’t look at it
When we prompted it to reference our purchase history, Rufus told us to check it ourselves.
One of the biggest value Amazon Rufus can provide is to reference your previous orders to recommend products. Right now, it doesn’t. And that’s a missed opportunity.
Rufus chats but doesn’t store conversations
We closed the window by accident and when we opened a new one, our conversation was nowhere to be found.
It’s quite a bummer because we had to redo the entire shopping process.
It also means we won’t be able to remember why we bought certain items.
When we mentioned we’re new to coffee and asked it where we can start, instead of educating us about coffee making, it went straight into recommending products. Much like an overly eager salesperson.
Comparing Amazon Rufus to ChatGPT Shopping and Perplexity Shop
AI shopping agents are becoming more prevalent.
They give ecommerce a different kind of edge. After all, who’s going to ask an in-store assistant, “What’s the best air fryer under $100 that’s good for small kitchens?”And how many reviews would a sales associate have read to give you a truly objective answer?
Amazon Rufus came first, then Perplexity Shop, and most recently, ChatGPT Shopping.
Here’s a factual comparison of all three
With Amazon’s vast amount of first-party shopping data, Rufus should have a huge advantage.
It knows how many people bought each item, when they bought it, what they returned, and likely even which demographics prefer which products.
But it doesn’t feel like Rufus is tapping into any of that. And that’s a pity.
Subjective impressions
In our testing, ChatGPT and Perplexity consistently understood us better.
Even when we didn’t phrase things perfectly, or skipped some context, they get it.
Rufus, on the other hand, often needed us to repeat or rephrase. Even with that, it sometimes misses the point.
Secondly, ChatGPT also has the advantage of memory (when enabled), allowing it to personalize recommendations over time.
Unlike Rufus, which treats every query as a blank slate, ChatGPT can remember previous preferences or needs you've shared—like budget, use case, or the fact that you already bought a specific product.
This lets it build continuity across sessions, making the experience feel more like an ongoing conversation than a one-off search.
Most importantly, ChatGPT and Perplexity provide far more explanation and context to why they’d recommend certain products.
They don’t just list products — they break down why you might want one over another, what trade-offs to consider, and even when a product might not be right for you.
Recommendations also come with summaries of reviews, pros and cons, and links to additional sources. This level of transparency gives shoppers more confidence to buy, especially if they’re unfamiliar with the category.
What this means for ecommerce brands
So yes, Rufus still has a long way to go.
But ecommerce brands should already be factoring it into your strategy. It’s pretty clear where things are headed: shoppers asking questions, AI answering in real time, and your product content being what the model pulls from.
If you’re already investing in Amazon SEO, this is just the next layer. It’s not about stuffing keywords — it’s about making your listing easy to understand, helpful to shoppers, and structured in a way LLMs can process.
Rufus might not be how most shoppers are discovering products yet, but the groundwork you lay now puts you in a better spot as it improves over time.
