Bye Rufus, Hello Alexa: Amazon's Big Bet on AI Shopping

Bye Rufus, Hello Alexa: Amazon's Big Bet on AI Shopping

Amazon has quietly retired the Rufus product name and folded it into a new offering: Alexa for Shopping. In this episode, Alasdair and Cameron break down what the rebrand really signals, why Rufus's reported 300M+ transactions in 2025 are being absorbed rather than abandoned, and whether Alexa can realistically compete with ChatGPT, Claude, and Gemini for the future of agentic commerce. They dig into Amazon's inherent bias problem (an Alexa shopping agent will never recommend Walmart), the $50B+ ad revenue at stake if consumers shift to LLM-first product discovery, the rumored Alexa phone, and why Prime's delivery moat still keeps Amazon untouchable — for now. Plus: shopping homework for next episode. Chapters 00:00  Cold open 00:42  Amazon axes the Rufus name 01:45  Why fold Rufus into Alexa? 04:16  The Alexa Plus push and unlocking voice purchasing 05:09  Alexa vs. ChatGPT voice — the capability gap 06:40  Inside Amazon's official announcement 07:13  Is Alexa biased? Best result for you vs. best for Amazon 10:39  Prime shipping moat vs. price-driven LLM picks 12:06  ChatGPT ads and using LLMs for real product research 14:19  Will Amazon ever unblock the LLMs? The TikTok parallel 15:42  The $50B ad revenue problem 17:34  Amazon's hedging strategy and the rumored Alexa phone 19:42  Where do consumers shop first — Amazon or LLMs? 21:13  Why Prime's moat still holds 22:14  The discovery spectrum: TikTok, LLMs, and agentic commerce 23:43  Homework: shopping experiments for next episode Key Takeaways * Rufus is being absorbed, not killed. Amazon is folding Rufus's functionality and reported 300M+ 2025 transactions into the marquee Alexa brand rather than retiring the product itself. * It's a branding consolidation play. One AI voice, one ecosystem — fewer competing personas, more weight behind Alexa Plus. * Convenience vs. trust is the real split. Alexa for Shopping is the convenience layer; ChatGPT and Claude are increasingly the "trusted advisor" for real product research. * The bias problem is structural. Alexa for Shopping will never recommend a Walmart deal, which limits its credibility as an unbiased agent. * Amazon's $50B+ ad business is the elephant in the room. If consumers shift from keyword search to LLM-first product discovery, the entire sponsored-results revenue model is at risk. * Amazon is almost certainly hedging. Big investment in OpenAI, rumored Alexa phone, job postings for agentic commerce — multiple bets running in parallel. * Prime's moat is still untouchable. Two-day (and same-day) shipping, easy returns, and delivery infrastructure remain Amazon's strongest defense. * Discovery is becoming a spectrum. TikTok-style algorithmic discovery at the top, LLM-driven decision-making in the middle, and Amazon convenience at the bottom of the funnel. * Watch what consumers do next. If shoppers start opening an LLM before Amazon, the balance of power shifts fast.

Announcing Market IQ: AI-Enabled Competitive Intelligence for Amazon & Walmart

Marketplace competition on Amazon and Walmart is a 24/7 problem — and most sellers don't have the time, data, or tooling to stay ahead of it. In this episode of HITL, we announce the launch of Market IQ, an AI-enabled competitive intelligence engine designed to tell sellers exactly who they're competing against, where their ad dollars are going, and what to do next to win page-one rank. We walk through the core problem Market IQ solves, why page-one visibility drives roughly 89% of clicks and sales, and how the product distinguishes between true substitutes, adjacent competitors, and "auction inflators" creating noise. We also dig into how Market IQ uses 13+ signals — including pricing, reviews, inventory, and listings — to surface insights that generic AI tools and ChatGPT-style lookups simply can't match. Finally, we cover what's coming next: in-app workflows launching in the next 60 days that will let sellers take action directly inside the product, plus how Market IQ fits into the broader Teikametrics ecosystem as the "navigation compass" for advertising and catalog decisions. 2. Key Takeaways * Competition on Amazon and Walmart is a 24/7 challenge — new sellers, new pricing, and new ad strategies emerge constantly, and human teams can't track it all manually. * Page one is where ~89% of clicks and sales happen, which is why Market IQ is laser-focused on page-one visibility for both paid and organic results. * Market IQ is built on real ad-spend data, not generic competitor lookups — it shows you exactly who's competing against you on the keywords you're already investing in. * The product uses 13+ signals (pricing, reviews, inventory, listings, and more) to classify competitors into meaningful buckets: direct substitutes, adjacents, and auction inflators. * Not all competitors deserve the same response — Market IQ helps sellers decide when to play defense, when to go on the attack, and when to ignore noise. * Market IQ is the connective tissue of the Teikametrics platform — feeding insights into advertising, catalog (Smart Pages), and listing optimization workflows. * It's available now for advanced and enterprise customers, with in-app action workflows shipping in the next ~60 days. * The earlier you plug in, the better — Market IQ compounds in value as the AI continues to learn your category and competitive set. Chapters: 00:00 Cold Open: Why Sellers Can't Keep Up With the Competition 00:46 The Biggest Problem Sellers Face on Amazon & Walmart Today 01:32 What Is Market IQ? 02:00 Why Page One Is the Real Battleground 02:18 How Market IQ Works 03:24 Why 89% of Clicks & Sales Happen on Page One 03:51 What Sets Market IQ Apart 05:07 The Three Types of Competitors 05:47 Market IQ + the Teikametrics Platform 07:00 Who Market IQ Is Built For 07:51 What's Next: In-App Action Workflows 08:41 Why Plugging In Early Matters 09:08 How to Get Access Join our free in-depth Market IQ webinar at the end of this month for a full product walkthrough with live visuals. The link is in the description below — register now to see Market IQ in action.

How to Break Through on TikTok: Lessons for Big Spenders & Breakthrough Brands

Is TikTok really for every brand? Ali Tatarzyn, Director of Product at Teikametrics, breaks down what's working — and what's not — for brands on TikTok in 2026. TikTok has officially moved past the experimental phase. Brands are allocating real budgets, building dedicated teams, and treating it as a serious growth channel — but most are still getting stuck. In this episode, Ali Tatarzyn, Director of Product at Teikametrics, breaks down what she's hearing from brands every week, from enterprise players investing heavily to smaller brands just dipping a toe in. She shares why TikTok isn't a fit for every brand, the two very different conversations happening across the industry right now, and the mindset shift required to actually win on the platform. Whether you're trying to prove halo impact to leadership or just trying to get your first piece of content to land, this conversation lays out where brands get stuck — and how to break through. Key Takeaways TikTok isn't for every brand — it rewards a specific way of operating that's fast, content-heavy, and creator-driven, and brands unwilling to adapt will struggle. The real challenge isn't ad spend or optimization, it's content. Most brands aren't built for the speed and volume TikTok demands. Organic, affiliate, and paid aren't separate strategies — they form a loop. Test organically, scale with affiliates, amplify winners with paid. Two distinct conversations are happening right now: mature brands focused on halo impact and measurement, and breakthrough brands trying to figure out why nothing is sticking. Big spenders get stuck on attribution and proving incrementality to leadership. Breakthrough brands get stuck producing content that's too polished and too brand-heavy. Letting go of creative control is one of the hardest — and most important — shifts brands need to make. Trust your creators to do what they do best. The time to invest is now. With Ulta, Sally Beauty, and other major retailers joining TikTok Shop, the platform is only getting more serious — but it's not too late. Don't bring your Amazon or Meta playbook to TikTok. Speed, volume, and creative iteration win over structure and control. Chapters 00:00 — Intro: Why Ali has a front-row seat to the TikTok conversation 00:54 — Is TikTok for every brand? A spicy take 03:20 — The biggest questions brands are asking right now 06:00 — The organic, affiliate, and paid loop 09:02 — The two camps of brands on TikTok 11:00 — Where big-investment brands get stuck (and how to fix it) 14:08 — The halo effect and the measurement problem 15:33 — Where breakthrough brands get stuck 17:22 — Why relinquishing creative control matters 18:38 — Final takeaways: the time is now 21:45 — Ulta, Sally Beauty, and what big retail moves signal for TikTok

TikTok Halo: Discover on TikTok, Buy on Amazon

The dominant consumer pattern of 2026 isn't buy on TikTok — it's discover on TikTok, buy on Amazon . That's great for the brands paying attention. It's a measurement nightmare for everyone else. In this episode, Cameron Yoder is joined by Alasdair McLean-Foreman, CEO and Founder of Teikametrics to break down the TikTok Halo: Why traditional attribution falls apart between the two platforms, why the discounting strategy that works for native TikTok Shop products actively hurts established brands, and how creator behavior — including off-message claims like "this product is being discontinued" — turns into a brand-safety problem the moment a video goes viral. The conversation gets practical: what big brands should actually measure (hint: not direct ROAS), why TikTok belongs in an experimental, upper-funnel bucket for most established brands, why the content engine is the muscle to start building today even if you can't justify the spend yet, what Amazon tried and abandoned on the discovery side, and why the DTC brands that swore off Amazon five years ago are quietly leaning back in — using Amazon Prime as a fulfillment layer rather than a competing channel. Takeaways at the end, and a tease that this won't be the last TikTok Halo conversation on the show. Timestamps / Chapters 00:00 — Cold open: discover on TikTok, buy on Amazon Prime 00:44 — What "TikTok Halo" actually means: the attribution gap, defined 01:55 — Why discounting blows up the strategy for established brands 03:39 — Consumer trust: Amazon Prime vs. TikTok, and the younger-audience split 04:46 — When creators go off-script: the beauty brand and pet food cases 07:00 — The metrics question: why direct ROAS will disappoint you 09:21 — The right framing: experimental bucket, upper-funnel marketing 11:36 — Build the content muscle now (and why Amazon's TikTok-style feed failed) 15:00 — DTC's quiet pivot: leaning back into Amazon as a fulfillment layer 16:28 — Key takeaways and what's next Key Topics / Talking Points What you'll learn in this episode: What the TikTok Halo is and why it's the dominant cross-channel pattern in 2026 Why "discover on TikTok, buy on Amazon" is the strategy big brands need to plan around — even if it's harder to measure Why the discounting playbook that works for native TikTok Shop brands actively damages established brands' merchandising How consumer trust differs between TikTok and Amazon, and where the younger-audience exception applies Real examples of creator-driven brand risk (e.g., false discontinuation claims) and how to think about brand control on a content-first platform The metrics big brands should and shouldn't use to evaluate TikTok performance Why TikTok belongs in an "experimental bucket" — and what that bucket should fund Why building the content engine today matters more than nailing the ROAS math What Amazon tried on the discovery side, why it didn't work, and what that says about platform moat Subscribe on Spotify, Apple Podcasts, YouTube and beyond for more AI-related content!

What the hell is MCP — and can it actually run your marketplace business?

Alasdair and Cameron debate whether you can really plug Claude into Amazon, let it rip on a $500K/month ad budget, and call it a business. Episode Summary: Everyone's talking about MCP — Model Context Protocol — and the dream of plugging a large language model like Claude straight into Amazon to run an entire marketplace business. In this episode, Cameron Yoder and Alasdair McLean-Foreman cut through the LinkedIn buzz to define what MCP actually is (spoiler: it's effectively a glorified API), where it genuinely shines today, and where the wheels come off when you try to hand a $500,000/month ad budget to a model with no middle layer. They get into the specifics: which use cases are ready right now (analytics, querying, cross-channel data exploration), why running an enterprise marketplace business demands more than a pipe between an LLM and a database, and what an MCP-native operating model actually looks like — including how it might compress headcount, why you'd want to build your own data layer, which categories of SaaS are most exposed (HubSpot Breeze, Zapier, legal tools all come up), and the Nvidia executive's "every software company becomes a token factory" line. Key Topics / Talking Points What MCP (Model Context Protocol) actually is — and why "it's a glorified API" is the right mental model Why plugging Claude directly into a large Amazon brand is, at this moment, a stretch — and what specifically breaks The use cases that already work today: analytics, ad-hoc querying, replacing BI/spreadsheet workflows The use cases that don't work yet: autonomous bid management, full-budget campaign edits, cross-channel optimization Why a "middle layer" — your own database and interface — is non-negotiable for serious marketplace operators How MCP changes headcount math: not zero people, but maybe one person doing three jobs The Anthropic "one-person marketing team" precedent and what it does and doesn't prove The Nvidia "token factory" thesis and which SaaS categories are most exposed Why HubSpot Breeze and Zapier hint at where mainstream B2B software is heading The real competitive threat to incumbent marketplace SaaS: lean, AI-native, agentic-from-day-one startups Why the right move for most operators is to test MCP — not bet the business on it Timestamps / Chapters 00:00 — Cold open & opening positions: can you really plug Claude into Amazon? 01:33 — What MCP actually is: the "glorified API" definition 03:42 — The "right" use cases: where Claude shines vs. where it breaks 05:11 — Would you let it run a $500K/mo ad budget? The middle-layer debate 07:04 — The data gap: cross-channel signals and why a pipe isn't enough 09:28 — The "token factory" thesis and headcount implications 11:17 — Architecting it yourself: which SaaS categories get displaced 14:04 — HubSpot Breeze, Zapier, and the AI-native competitor threat 17:17 — Takeaways and the buy-a-brand experiment tease If you got value from this episode: * Subscribe to Human in the Loop on Spotify and YouTube so the next episode lands in your feed. * Follow Cameron and Alasdair on LinkedIn — that's where the between-episode debates and experiment updates live. * Try the test yourself. Spin up an MCP connection on a small slice of your business and see what it can actually do — you don't have to bet the budget to learn something. * Tell us what to sell. We're seriously considering buying a brand and running it with AI in the loop. Reply on LinkedIn or email the show with what category we should pick. * Rate the show wherever you're listening — it's the single biggest thing that helps new operators find us.