On May 13, Amazon quietly killed Rufus. Not really: they rebranded it.

Rufus is now Alexa for Shopping, and the change rolling out across Amazon, the mobile app, and Echo devices over the coming week is bigger than a name swap.

It's Amazon's answer to the AI shopping threat that's been chewing into their search bar dominance.

Scot Wingo and Christian Umbach both dove deep on what sellers need to know.

What Actually Changed

Amazon merged its two AI assistants, Rufus (shopping) and Alexa+ (everything else), into one unified assistant called Alexa for Shopping. The Rufus icon is being replaced by an Alexa icon everywhere a customer touches Amazon. No Echo device, no Alexa app, no Prime membership required. Every signed-in Amazon customer gets it for free.

The Rufus numbers Amazon dropped to justify the move: 300 million+ customers used Rufus in 2025, it drove $12 billion in incremental annualized sales, and shoppers using it were 60% more likely to complete a purchase.

Rufus is out of beta. This is now the front door.

The Eight Features Sellers Should Care About

That last phrase — across Amazon and the web — matters. Combined with Buy for Me (launched April 2025, uses agentic AI to buy products from third-party sites without the customer leaving the Amazon app), Amazon is no longer just a marketplace. It's a shopping agent that will buy off-Amazon products on the customer's behalf.

Why Amazon Did This Now

For over half a decade, Amazon's search bar was the default starting point for product searches. 50%+ of all product searches in 2023 started there, versus 31% on Google. That position is now under siege from a direction Amazon didn't have to defend against before: conversational AI.

63% of consumers say they plan to use AI chatbots for shopping in 2026. 70% of people who've tried ChatGPT for product recommendations prefer it to traditional search. Every "what's the best [X] for [my situation]" question answered by ChatGPT is a query that never reaches Amazon's search bar.

The bigger threat isn't ChatGPT recommending products. It's the horizontal context ChatGPT is collecting. People tell ChatGPT about their trips, their kids, their health, their finances, their meal plans. Five turns into a skincare conversation, ChatGPT recommends a product Amazon doesn't even sell.

Amazon is blind to all of that adjacent context. Alexa for Shopping, with persistent cross-device memory and an expanding remit into health, travel, and home services, is Amazon's attempt to capture that context on their own turf before someone else does.

What This Means for Your Listings

The AI layer between your listing and the shopper is now permanent. Rufus was easy to dismiss as an experiment. Alexa, with 600M+ active endpoints and Amazon's most recognized AI brand behind it, is not.

Your listing content, A+ content, reviews, and structured data are now feeding the AI that decides whether to surface your product in a recommendation, comparison, or shopping guide.

Keyword-only listings are going to lose ground. Shoppers aren't typing "yoga mat" as much anymore. They're asking "what's the best yoga mat for a beginner with knee problems." Listings that answer questions, address specific use cases, and provide context the AI can actually use will get pulled into comparisons and shopping guides. Listings that just stack keywords won't.

AEO is the new SEO inside Amazon. Title, bullets, A+, backend, reviews, every one of these is now also training data for an AI that's making the recommendation. The brands that win the next 12 months will be the ones whose listings read like answers to questions, not like keyword soup.

The compare-products feature is a double-edged sword. If your listing wins comparisons on features, price, and review quality, you get free placement next to your competitors. If you lose on any of those dimensions, the AI is now explicitly showing the shopper why your competitor is better.

Price history is now native. A customer can see your last 12 months of pricing with one tap. Aggressive coupon stacking and "fake" sale prices have a shorter shelf life than ever. Plan promotions like the customer can see what you charged last August, because now they can.

Scheduled actions change replenishment. If a customer sets a scheduled action for "kids' snacks every month" and Alexa picks the product, you want to be the product Alexa picks. That decision will increasingly be made on the strength of past purchase data, reviews, AI-overview quality, and listing content, not just keyword rank.

Amazon renaming Rufus to Alexa for Shopping is the same kind of signal Google sent when it folded everything into Gemini. The experimental phase is over.

This is how shopping works now.

Andy Jassy said it in his shareholder letter last month: "When transformative technology like generative AI arrives, and you can build a much more intelligent product than you previously had, you have to pursue it, even if it's disruptive to your team, roadmap, and architecture."

Translation for sellers: the rules of how your product gets found, compared, and chosen on Amazon are being rewritten in real time. The brands that adapt their listings, content, and pricing for an AI-mediated shopping experience are going to take share from the brands that keep optimizing like it's 2025.

Rufus walked so Alexa for Shopping could run. Optimize accordingly.

Most Amazon sellers think going to Shopify is just uploading their listings to a new platform. That thinking is exactly what's wrecking brands before they even get out of the gate.

This week on Marketing Misfits, Norm and Kevin chat with Kurt Elster, the host of the Unofficial Shopify Podcast and one of the sharpest Shopify strategists in e-commerce. And he doesn't pull punches.

Kurt breaks down the real gap between demand capture (Amazon) and demand creation (Shopify), why free shipping is way less powerful than you've been told, and what the 7- and 8-figure Shopify brands are doing right now that 95% of sellers are completely ignoring.

We also dig into AI-powered product discovery, identity resolution tools that can pull 40–70% of your "anonymous" traffic out of the shadows, TikTok Shop, and the brand-new Universal Commerce Protocol. It’s the thing that's about to decide whether ChatGPT recommends your products or your competitor's.

If you're an Amazon seller eyeing omni-channel, a Shopify operator getting crushed by rising CAC, or just trying to figure out where e-commerce actually goes in 2026, this one's required listening.

🔑 What you'll walk away with:

→ Why most Amazon sellers build "thin" Shopify stores — and the fix
→ The #1 driver of Shopify traffic in 2026 (hint: it's still Meta)
→ How identity resolution captures 40–70% of anonymous site visitors
→ Why free shipping isn't moving the needle — and what actually does
→ The "free gift with purchase" play that lifts AOV and conversion rate
→ How Shopify's UCP gets your products surfaced by ChatGPT

🎧 Listen here 📬 Get the free Marketing Misfits newsletter: https://misfits.news

If you've been running an Amazon business for the last decade, you've watched your toolstack balloon. Helium 10 for keywords. Data Dive for reverse ASIN. Jungle Scout for product research. A separate PPC bidder. A separate review aggregator. Each one a dashboard. Each one a monthly fee. Each one a tab you forget to close.

That entire model is now on the clock. And the thing killing it has a clunky technical name: MCP.

What Is an MCP, Actually?

MCP stands for Model Context Protocol. Anthropic introduced it, but it's now an open standard the whole AI industry is adopting. OpenAI, Google, and others have all built around it.

The simplest way to think about it: MCP is USB-C for AI.

Before USB-C, every device needed its own cable. Apple here, micro-USB there, weird proprietary plug for that one camera. USB-C made all of that go away. One plug. Works on everything.

That's what MCP does for AI agents. Before MCP, if you wanted ChatGPT or Claude to actually do something in your business, like pull data from Seller Central, send an email, or update inventory, somebody had to custom-build that connection for every tool and every AI. MCP makes the plug universal.

Your AI doesn't just know things. It can now do things, across every system in your business, without somebody wiring it up custom.

How It Actually Works

An MCP server is the worker's hands and eyes for one specific tool. There's one for Gmail. One for Seller Central data. One for your inventory system. Give your AI access to those servers, and it can read the data, take actions, and return results in conversation.

The kicker: the AI reasons across all those tools at once. You can say "pull my last 30 days of orders, cross-reference with my ad spend, find the SKUs where ACOS spiked, and draft a reply to my agency" and it works. One conversation. Five tools. No tab-switching.

That's the part that makes traditional SaaS dashboards start to look like fax machines.

Why Amazon Tools Have a Problem

Most of these tools' actual value isn't the data. It's the interface to the data.

Helium 10 doesn't have magic keyword data nobody else has. It scrapes Amazon, runs estimates, and packages it in a usable dashboard. Data Dive does sophisticated reverse-ASIN analysis, but the underlying data is also derived from Amazon. Jungle Scout, Carbon6, Sellerboard, Pacvue: same structural story. The moat is the workflow and UX.

MCP and agentic AI attack that moat from two directions.

From below: Amazon is taking back its own data. Product Opportunity Explorer has quietly become a serious research workspace. Saved Opportunities. An Unmet Demand report showing keywords where search volume surges but conversion lags, straight from Amazon's actual shopper behavior, not third-party estimates.

An Insights & Trends tab comparing today versus 90 and 360 days ago. The capabilities sellers used to pay Helium 10, Data Dive, and Jungle Scout to approximate are increasingly available at the source, for free, with first-party accuracy.

From above: AI agents are taking over the interface. Even if a third-party tool has the best data, who's going to log into a dashboard when their AI agent can query the data, cross-reference five other systems, and hand them an answer in a sentence? The dashboard used to be where the value lived. The conversation is where it's moving.

A tool that's only a dashboard is a tool that's about to be a feature inside someone else's agent.

The Titan Network Example: What Adaptation Looks Like

The clearest live example came out of a roundtable at Ecom Mastery AI in Nashville. Dan Ashburn, co-founder of Titan Network, a community and software platform 7+ years deep in the Amazon space, walked through what they're now calling Titan Connect.

Titan Connect is open MCP infrastructure. It takes Titan AI (their enterprise-grade knowledge system, trained on years of community content, refreshed every 24 hours into Microsoft's vector database) and lets you install it as a skill inside Claude or other AI agents.

Dan's line was the clearest articulation of the shift I've heard: "Claude on his own is like a college graduate, knows a lot about nothing and has never done anything. The goal with Titan Connect is to make Claude operate at a senior operator PH.D level."

Read that twice. They're not building a dashboard. They're building a skill that plugs into your AI, so your AI knows how a senior experienced Amazon operator thinks: pricing, PPC, launch sequencing, supplier negotiation, and applies that thinking to your specific situation.

That's the template. Not "here is our software, please log in." Instead: here is our expertise, expressed as an MCP-installable skill that makes your AI smarter.

What This Means for Helium 10, Data Dive, and Everyone Else

Every major Amazon SaaS company is staring at the same fork.

Option one: Become an MCP-native intelligence layer. Expose your data and proprietary methodologies through MCP servers. Stop competing for the seller's attention with a dashboard. Start competing for the AI agent's attention with the best data feed and the most useful operator-level reasoning.

Helium 10 already has structural advantages: they bundle education (Freedom Ticket) with software, plus a massive community. The path is to express all of that as agent-installable intelligence, not as another browser tab.

Option two: Get absorbed into someone else's agent. Point-solution tools, single-feature Chrome extensions, standalone PPC bid managers, listing optimizers end up as commodity features inside a more capable AI agent. Some sell themselves as acquisitions. Most just lose customers until the lights go off.

Option three: Pivot to thought leadership and services. Don't be a software company that happens to talk. Be a thought-leadership media company that happens to ship software. HubSpot won inbound marketing by giving away the course five years before anybody logged into the product. Helium 10 Elite runs a version of the same playbook in our space.

Data Dive is an interesting case to watch. Their proprietary reverse-ASIN methodology is genuinely sophisticated, exactly the kind of analytical workflow that would translate well to an MCP-installable skill.

And they are getting ready to make the move. Sitting on the dashboard model is a slow-motion problem when the same seller's AI agent can query Amazon's first-party data, cross-reference it with their own sales history, and deliver the same insight without ever opening a third-party app.

The companies that survive will be the ones who stop thinking of themselves as software and start thinking of themselves as operator-level expertise that an AI agent can install.

Jo's Take: AI Is a Strategist, Not a Tool

The mindset shift sellers themselves need to make is just as big. Jo Lambadjieva, who runs the AI for E-commerce newsletter and consults brands on AI integration, has been making this point loudly: forget prompt engineering hacks. The real unlock with AI is a fundamental shift in how you think, not what you type. Stop using AI like a tool. Start using it like a strategist.

Most sellers are still treating AI like a fancier search bar. Paste in a question, copy out an answer, paste it into their listing, move on. That's tool-level thinking. It misses 95% of the leverage.

Strategist-level thinking is different. It's saying: here is my brand, here is my customer, here is my data, here are my goals — now help me reason about what to do next. It's giving the AI the context to think alongside you, then letting it execute across every system once it knows the strategy. That's only possible when your AI can actually reach into your tools, which is exactly what MCP enables.

Jo's data-driven approach of mining real reviews for what buyers actually care about, translating Reddit and Pinterest signals into image briefs, building AI avatars for the Q&A questions Rufus is now gating gets dramatically more powerful when the AI has live MCP access to your account data instead of you copy-pasting CSVs back and forth.

Where This Leaves You

If you're a seller, stop evaluating tools the old way. Don't ask "what dashboard has the prettiest UI." Ask "does this thing have an MCP server or a clear roadmap to one, so my AI agent can actually use it?" Tools that can't answer that in twelve months will be tools you're paying for and not opening.

If you're running an Amazon SaaS company, the question is sharper. The dashboard is no longer the product. The intelligence is. Either you become the operator-level expertise that lives inside your customer's AI, or you become an icon they stop clicking.

The era of tab-switching between fifteen Amazon tools is ending. The era of one AI agent, with operator-level skills installed, reasoning across your whole business in plain English era has already started. The sellers and companies that figure out which side of the MCP line they're on first are the ones who win the next five years.

While Amazon sellers are still figuring out how Rufus Alexa decides what to surface, Shopify is busy turning every merchant into an AI-native operator, whether they asked for it or not.

The company just rolled out connector apps for both ChatGPT and Claude that let merchants run their entire store from inside the AI assistant.

Look up orders, update prices, check how a new collection is performing, upload a product photo and have the AI add it to the catalog, push discounts across a category, all from a chat window on your phone. No dashboard, no logging into the admin, no clicking through six menus to change a price.

This follows last month's Shopify AI Toolkit, which was aimed at developers using agents like Claude Code and Cursor. The new connectors are the merchant version built for the person actually running the store, not the dev building on top of it.

And then there's the quieter move that matters more long-term.

Shopify has started rolling out native llms.txt files on stores without any formal announcement (spotted by Anton Ekström). If you point a browser at yourstore.com/llms.txt on a Shopify store, you'll find store metadata (currency, contact info), direct links to product listings and search, agent instructions through dedicated endpoints, and MCP endpoints for programmatic commerce.

In other words, Shopify is handing the LLMs a structured map of every store on the platform, here's what we sell, here's how to query it, here's how to transact.

For Amazon sellers, the read here isn't "switch to Shopify." It's that the platform you sell on is about to start being judged by how well it feeds the agents. Shopify is making its merchants discoverable, queryable, and transactable by AI by default.

Amazon has Rufus Alexa and is building toward agentic commerce, but the surface area an Amazon seller can directly influence is still narrow: your listing copy, your A+ content, your reviews. Shopify is giving merchants structured hooks the LLMs can grab onto.

The funniest detail: the llms.txt file also includes a direct ad for Shopify and a call to action to start your own store. Aimed at the LLMs, apparently. Pitching the agents on becoming customers. We're already there.

The takeaway for Amazon sellers running a DTC site alongside their FBA business: your Shopify store may already have an llms.txt file you didn't know about. Worth pulling it up and looking at what it's telling the agents about your brand, because that's now part of your AEO surface whether you set it up or not.

✌🏼 Have a great weekend.

See you again on Monday.

The answer to today’s STUMP BEZOS is
Amazon says FBA is 70% cheaper than other options