They are most useful for research-backed analysis of AI shopping agents, marketplace strategy, and emerging platforms for ecommerce operators.
Best for
B2B ecommerce analytics campaigns that need analyst commentary for operators and marketplace sellers.
Not for
Paid consumer creator work, live-selling hosting, or short-form product video for fashion and beauty brands.
Key takeaways
AI shopping agents favor cheaper, higher-rated products and avoid listings marked Sponsored.
ChatGPT may have enough retail supply to focus on shopping experience over partners.
Whatnot's reported growth suggests live commerce may take hold in the West.
Check first: Confirm commercial terms, audience demographics, and reach or engagement on comparable posts.
Main ideas
AI shopping agents have their own biases, and paid placement can work against you · ChatGPT's retail strategy is a supply-and-demand problem, and brands shouldn't wait for a direct integration · Big shopping platforms can scale while staying invisible to the industry, so judge them on metrics rather than your own feed
Juozas Kaziukėnas writes about ecommerce, online marketplaces, and AI shopping.
Their posts examine how buying behavior is changing: how AI agents choose products, how retail partnerships affect what shoppers can find, and how a platform such as Whatnot can grow outside the notice of mainstream observers.
The posts offer concrete arguments about emerging shopping trends rather than broad predictions about technology.
Kaziukėnas connects specific examples and platform details to clear, sometimes counterintuitive takeaways.
In one post, they describe shopping agents favoring cheaper, higher-rated products and avoiding “Sponsored” listings, then argue that sellers should pay attention to those biases.
In another, they reason that ChatGPT may have enough retail supply to shift attention from adding partners to improving the shopping experience.
Their Whatnot post offers a check on their own assumptions: after arguing for years that live commerce would not take hold in the West, they treat Whatnot’s reported growth as a sign that they were wrong.
That mix of practical analysis, pointed predictions, and willingness to revise a view gives readers specific claims to examine—and a reason to follow how their assessments change.
Juozas Kaziukėnas writes about ecommerce and online marketplaces.
Read more
These posts track shifts in shopping behavior and platform power before most brands notice — from AI shopping agents and ChatGPT's retail partnerships to live commerce on Whatnot. A consistent method runs through them: ground claims in numbers or a specific study, separate business scale from personal taste, and end with a practical implication for sellers.
AI shopping agents have their own biases, and paid placement can work against you
Kaziukėnas treats AI agents as decision-makers with preferences, not neutral search indexes.
Read more
In a research paper where Claude, ChatGPT and Gemini were asked to make shopping choices, the agents tended to pick cheaper, higher-rated products and favored listings tagged "Overall Pick" or "Best Seller"; they were inconsistent with one another, but all avoided products tagged "Sponsored," even when those ranked at the top.
The bias appeared without any instruction about sponsorship, which is what makes it notable to Kaziukėnas.
The same paper found that AI-generated tweaks to product descriptions could sharply raise clicks.
Kaziukėnas frames this as the early stage of optimization for AI agents — a differentiating discipline where the rules are still being discovered and can currently be exploited with prompt injection — and argues the sponsored bias is something sellers could learn to align with rather than accept as fixed.
The test prompt asked agents only to “use your best judgment about what would work well for a typical person, and select one product to purchase” — sponsored labels were never mentioned, yet every engine avoided them.
ChatGPT's retail strategy is a supply-and-demand problem, and brands shouldn't wait for a direct integration
Kaziukėnas argues ChatGPT is not racing to sign up more retailers because three partners — Etsy, Shopify and Walmart — already give it broad coverage of ecommerce inventory, leaving gaps in electronics, fast fashion, furniture and luxury.
Read more
The rollout has been slow: four weeks after the Etsy and Shopify announcement, only Etsy was live.
They expect Walmart to take even longer, while eBay would join as soon as OpenAI called and Amazon would likely hold out.
Because ChatGPT already has demand, they expect the next phase to focus on scaling user behavior and building richer shopping experiences rather than adding supply — a point they sharpen by calling ChatGPT “very dull for shopping.”
Their conclusion for brands is concrete: sell on Walmart or host on Shopify now instead of waiting for direct integration.
Weeks after the partnership announcements, only Etsy was live — evidence Kaziukėnas cites to argue that OpenAI feels no urgency to expand its retailer list.
Big shopping platforms can scale while staying invisible to the industry, so judge them on metrics rather than your own feed
Kaziukėnas compares Whatnot to SHEIN a few years ago: billions of dollars in sales with almost no visibility outside its core audience.
Read more
They cite roughly $8 billion in live ecommerce sales in a year, 100%+ growth, more than half the US market in its categories, and months as one of the most downloaded US shopping apps behind only Temu, with users reportedly spending about 95 minutes a day in the app.
Its core categories — collectibles such as sports cards, trading card games, coins and entertainment cards, plus fashion and sneakers — are why general audiences miss it.
They are blunt that they find the content “low quality and boring,” a reaction they consider more interesting than disqualifying because the audience clearly responds to it.
Fashion and beauty are the fastest-growing categories, and they expect Whatnot to overtake Etsy in gross merchandise volume.
Kaziukėnas personally dislikes the content on a platform where users spend about 95 minutes a day — a gap between their taste and audience behavior that informs their analysis.
Kaziukėnas says they argued for years that China's live-commerce boom would not translate to the West, and that earlier attempts like Amazon Live went nowhere. Whatnot changed their mind, but they keep the revision narrow: it is a success in a specific niche, not proof of broad Western demand for live shopping.
Answers the basic question of how AI shopping agents choose products.
Read more
It is the most concrete, testable piece in the set, making it a useful starting point before drawing conclusions about marketplace strategy.
Kaziukėnas walks through a research paper that had Claude, ChatGPT, and Gemini make shopping decisions.
The agents tended to pick cheaper, higher-rated products, favored listings tagged "Overall Pick" or "Best Seller," and were inconsistent with each other — but all avoided products tagged "Sponsored," even when those ranked at the top of search results.
The prompt did not mention sponsored tags; it only asked the agents to use their best judgment.
The paper also found that AI-generated rewrites of product descriptions could sharply increase clicks.
Kaziukėnas notes that this can currently be gamed through prompt injection.
Their closing point: agents are not neutral; they carry biases, including one against sponsored results that a seller could try to align around.
For brands planning a campaign
A research-backed webinar or article could turn this into a practical session on which listing signals influence an AI agent's pick — rating, price, badges, description wording — and which, like a sponsored label, work against it. For a B2B ecommerce analytics audience, that is a directly actionable agenda item, not a general AI trend talk.
Answers why ChatGPT's shopping integrations are moving the way they are and what a brand should do about it.
Read more
It is a useful second read because it connects AI agents to marketplace supply rather than treating them as a standalone topic.
Kaziukėnas argues that three retail partners already give ChatGPT broad coverage of ecommerce inventory, leaving gaps in electronics, fast fashion, furniture, and luxury that Walmart, Etsy, and Shopify do not fill.
They note that the rollout has been slow: only Etsy was live four weeks after the Etsy and Shopify announcement, with the Walmart launch expected to take even longer.
In their view, eBay would join as soon as OpenAI called, while Amazon would likely hold out.
Because ChatGPT already has demand, they expect OpenAI to focus on user behavior and richer shopping experiences rather than rushing to add more retailers.
Their advice to brands is to sell on Walmart or host on Shopify now instead of waiting for a direct integration.
For brands planning a campaign
This gives a brand a concrete near-term decision: where to list inventory to be reachable inside AI shopping surfaces. It also supports a contrarian webinar angle: the bottleneck is not retailer access but the shopping experience itself, which Kaziukėnas views as still dull.
Shows what a fast-growing but low-visibility shopping platform looks like before it becomes mainstream.
Read more
It pairs well with the AI-agent posts because it shows the same audience making a bet on where shopping behavior is actually moving.
Kaziukėnas compares Whatnot to SHEIN a few years ago: billions of dollars in sales with almost no visibility outside its core audience.
They cite $8 billion in live ecommerce sales the prior year, 100%+ growth, more than half of the US market in its categories, and months as one of the most downloaded shopping apps in the US, behind only Temu.
Its main categories are collectibles — sports cards, trading card games, coins and money, entertainment cards — and fashion, including clothing and sneakers.
They report users spending 95 minutes a day on the app.
Kaziukėnas is blunt that the content quality is low and boring, which they find interesting precisely because the audience clearly responds to it.
They predict Whatnot will likely overtake Etsy in gross merchandise volume and flag fashion and beauty as its fastest-growing categories.
They also note that they had argued for years that live commerce would not translate from China to the West, and say Whatnot is the case that changed their mind.
For brands planning a campaign
For a brand evaluating where to test live shopping, this is a reason to look at Whatnot before it becomes an obvious choice, particularly in fashion and beauty rather than only collectibles. The post also offers a useful analytic move: taking a platform seriously because its audience behaves differently from the reviewer's own taste, an approach a campaign brief can borrow.
How AI shopping agents pick products — and why "sponsored" labels backfire
Kaziukėnas walks through a research paper that tested how Claude, ChatGPT, and Gemini behave when asked to buy something.
Read more
The agents tended to choose cheaper, higher-rated products, favored items labeled "Overall Pick" or "Best Seller," and avoided products tagged "Sponsored" — even when sponsored results ranked at the top.
They notes the engines were not fully consistent with each other, but all shared the sponsored-label avoidance.
The same paper found that AI-generated tweaks to product descriptions could sharply increase clicks.
Kaziukėnas frames this as the early stage of optimization for AI agents, where the rules are still being discovered and can even be gamed with prompt injection.
Reads a specific research paper, summarizes its findings on agent behavior, then adds their own interpretation — including the idea that agent bias could be deliberately aligned or corrected rather than accepted as fixed.
For brands planning a campaign
Brands building listings or content for AI-mediated shopping should treat agent behavior as a distinct audience with its own biases, not as a neutral index.
Sponsored labeling, which normally buys placement, may actually reduce an agent’s likelihood of selecting a product.
This is a concrete hook for a webinar or article on preparing product pages for agentic commerce.
Why ChatGPT's retail partnerships change where brands should sell
Kaziukėnas argues ChatGPT is not racing to add more retailers.
Read more
Three partners — Etsy, Shopify, and Walmart — already give it broad coverage of ecommerce inventory, though with clear gaps in electronics, fast fashion, furniture, and luxury.
They note the rollout has been slow: weeks after announcements, only Etsy was live.
Their reading is that OpenAI has solved supply and demand through those integrations, so the next phase is scaling user behavior and improving the shopping experience rather than signing eBay or Amazon — eBay would join quickly if asked, while Amazon will likely hold out.
Uses a supply-and-demand frame to interpret partnership announcements and rollout timing, then translates it into a concrete listing decision for brands.
For brands planning a campaign
For brands weighing where to list, Kaziukėnas gives a direct answer: selling on Walmart or hosting on Shopify is the practical route into ChatGPT's shopping surface, and waiting for a direct integration is a mistake. This is a usable recommendation for a marketplace-strategy piece or webinar segment.
Whatnot and the arrival of live commerce in the West
Kaziukėnas profiles Whatnot as a live-commerce platform that has scaled largely out of view.
Read more
They cite roughly $8 billion in live ecommerce sales in a year, growth above 100%, more than half the US market in its category, and months as one of the most downloaded US shopping apps, trailing only Temu.
Its core categories — collectibles such as sports cards, trading card games, coins, and entertainment cards, plus fashion and sneakers — keep it invisible to general audiences.
Users reportedly spend about 95 minutes a day in the app.
They are candid that they find the content low quality and boring, which makes the platform's scale more interesting to them, and note fashion and beauty are the fastest-growing categories.
Combines platform metrics with their own firsthand impression of the content, and openly revises their earlier prediction about Western live commerce.
Kaziukėnas says live-commerce success in China did not guarantee Western adoption, and that earlier attempts like Amazon Live went nowhere — so Whatnot's rise is a specific niche success, not proof of broad Western demand for live shopping.
For brands planning a campaign
Live commerce is a distribution channel worth evaluating on reach and engagement rather than production polish, especially for collectibles, fashion, and beauty brands. Kaziukėnas also supplies a candid counterpoint: they spent years skeptical that China's live-commerce boom would translate to the West, and say Whatnot proved them wrong — useful context for a balanced assessment rather than a hype piece.
Juozas Kaziukėnas's posts disclose no sponsorship, affiliate, employment, ownership, client, or partnership relationship.
Read more
They analyze marketplaces and shopping technology, including OpenAI's ChatGPT commerce integrations, SHEIN and Whatnot as emerging marketplaces, and how AI agents treat sponsored search results. These companies appear as subjects of analysis, not stated business relationships; ask directly about any commercial ties.
Kaziukėnas writes short, argument-led posts about where online shopping is heading: AI agents that choose products, ChatGPT's retail partnerships and live commerce on Whatnot.
Read more
Start with the post on a research paper that tested Claude, ChatGPT and Gemini as shoppers.
It shows their recurring approach: take a specific study, pull out the finding that matters, then explain what it means for a seller.
The ChatGPT partnership post is most useful if you are deciding where to list a product.
The Whatnot post best shows how they can change their mind: after years of doubting that Western live commerce would take off, they treat the platform's numbers as proof they were wrong.
Read those three in that order for a mix of data, personal taste and seller-facing advice.
Then follow the tags below or compare Kaziukėnas with Laurie Kacinski, Sharon Gai and Haider Ali, who cover overlapping ground from different angles.
Writes about the same marketplaces from the inside, using specific brand numbers — 3,395 unauthorized sellers found for a health and wellness brand, third-party Walmart offers averaging 41% below a brand's everyday price — to show what is happening beneath a sales dashboard.
Read more
That data-first approach is close to how Kaziukėnas uses figures about ChatGPT's partners or Whatnot's sales to explain marketplace shifts.
Kacinski's posts are operational checklists for brand-protection and ecommerce teams, ending with an invitation to book a call. Kaziukėnas focuses on platform economics and shopper behaviour, with no service pitch in these posts.
Covers the same move toward AI agents doing the buying, from payment rails such as Google's AP2, Mastercard's Agent Pay and Visa's Trusted Agent Protocol to consumer habits like sending a face photo to ChatGPT for skincare advice.
Read more
Both treat agentic commerce as a real change in how purchases happen, not a slogan.
Gai explains infrastructure and consumer trends, framing the shift as an opportunity for 'agent-ready' retailers and closing with a wry note about handing a credit card to an agent. Kaziukėnas questions the structure behind the shift — why agents avoid sponsored labels, why OpenAI is not rushing to add retailers — and is blunt when they find something dull or inconsistent.
Works the same AI-and-ecommerce seam from a seller's side, covering agentic shopping protocols such as the Universal Commerce Protocol and explaining how to run Amazon competitor research with AI tools.
Read more
Both write for people who sell on marketplaces and need to know what to do next.
Ali publishes step-by-step setup instructions and copy-paste prompts, with 'are you ready?' urgency, and signs each post with a date and company. Kaziukėnas writes short analytical posts that end with a strategic conclusion rather than a workflow or tool setup.
For a campaign aimed at ecommerce-analytics buyers, the practical question is whether Juozas Kaziukėnas can put AI-agent buying behavior and marketplace strategy in front of that audience.
Read more
Their posts address that conversation: a 27 October 2025 post on ChatGPT’s retail partnerships and what they mean for brands selling on Walmart or Shopify; a 27 August 2025 post on AI agents avoiding “Sponsored” results; and a 29 January 2026 post on Whatnot’s live-commerce scale in collectibles and fashion.
The LinkedIn profile reports 14,322 followers as of 11 October 2026, but does not show their locations, job functions, seniority, employers, or buyer roles.
Treat topic fit as strong and professional-audience fit as unverified.
Ask for the creator’s most recent audience survey or platform audience demographics, along with recent reach or engagement on comparable posts about AI agents and marketplace strategy.
If no such source exists, ask how they know the audience includes ecommerce analytics buyers.
Juozas Kaziukėnas's three posts offer no sponsor policy, rate card, media kit or disclosed example of paid partnership work, so there is no basis for describing a partnership format.
Read more
They show the kind of output they produce: short analytical commentary on ecommerce, marketplace supply and AI shopping agents, written for a professional audience rather than as branded content.
For a B2B ecommerce analytics company, a realistic starting point is to approach them as an analyst or commentator whose posts could frame a webinar or article topic, not as a creator with published sponsorship placements.
No official partnership inquiry route is published.
Ask directly whether sponsored posts, research collaborations or paid speaking are available, and on what terms.
Strong fit, with the format grounded in what Juozas Kaziukėnas's posts show. They produce short, sourced analytical commentary for a professional audience. Each of the three posts starts with a concrete data point — a research paper on AI shopping agents, a partnership rollout, or a platform's sales growth — and follows with a recommendation. That is the shape a research-backed B2B session needs.
Read more
Brief
A B2B ecommerce analytics company wants to reach ecommerce operators and marketplace sellers with a research-backed webinar or article on how AI agents and emerging platforms like Whatnot reshape online shopping. The deliverable is analysis: a session or long piece that turns primary research and platform data into decisions sellers can act on.
Reasons
The topic match is direct, not adjacent. One post reads a research paper on how Claude, ChatGPT, and Gemini choose products: they favored cheaper, higher-rated items, favored "Overall Pick" and "Best Seller" labels, and avoided "Sponsored" tags even when those ranked first. That finding is the core agenda item for an ecommerce analytics audience, and it comes with a testable conclusion rather than a trend statement.
The marketplace-strategy angle is already framed as advice. On ChatGPT's retail partnerships, the post argues that integrations with Etsy, Shopify, and Walmart have solved supply, so brands should sell on Walmart or host on Shopify rather than wait for a direct integration. For an audience deciding where to list, that is a usable recommendation to examine in a webinar.
The Whatnot post supplies the second half of the brief. It covers roughly $8 billion in live ecommerce sales, growth above 100%, a claim of more than half the US market in its categories, about 95 minutes a day of user time, and fashion and beauty as the fastest-growing categories. It also records a reversed prediction: the post says the author had argued for years that live commerce would not translate from China to the West, and that Whatnot changed that view. That self-corrected data point can help make for a balanced session.
The audience signal is a LinkedIn account with 14,322 followers, observed on 11 October 2026. That supports professional-audience reach but does not establish whether followers are ecommerce analytics buyers. Ask for audience-role data before booking a lead-generating webinar.
Conditions
Treat the engagement as analyst or commentator work, not paid creator content. Juozas Kaziukėnas's posts do not specify a sponsor policy, rate card, media kit, or past sponsored placements. Agree on commercial terms directly before booking.
Ask for reach or engagement figures on posts comparable in depth to the proposed webinar, plus any audience survey or platform demographic data. Buyer-role fit is not established without them.
Build the session around the research paper discussed in the sponsored-tags post and the platform data in the partnership and Whatnot posts. Mark claims based on the creator's own reading, rather than published data, as interpretation.
When to choose a specialist
Reject for this brief based on what Juozas Kaziukėnas's posts show. The topic adjacency is misleading: the posts discuss fashion and beauty as Whatnot's fastest-growing categories, but commenting on a category does not demonstrate experience hosting live sales or producing consumer product content.
Read more
Brief
A fashion or beauty brand wants a paid creator partner to host shoppable live-selling segments and produce short-form product video that drives impulse purchases from a consumer audience.
Reasons
Juozas Kaziukėnas's posts do not describe a commercial format or specify a sponsor policy, pricing, media kit, or examples of paid partnership work. Ask directly for commercial terms and relevant examples before quoting or briefing a paid placement.
Juozas Kaziukėnas's posts show written commentary, not the requested deliverable format. They provide no evidence of on-camera presentation, live hosting, or short-form video production; a post about Whatnot's scale does not demonstrate Whatnot-style selling. Ask for relevant work samples before commissioning that format.
The audience is unverified for a consumer brand. The only measured audience signal is 14,322 LinkedIn followers as of 11 October 2026. The posts do not give geography, job function, seniority, employer, or buyer-role data. Ask for those audience details before committing to a consumer fashion campaign.
The tone runs against the brief. The Whatnot post describes the platform's content as low quality and boring while taking the audience's behaviour seriously. A brand expecting polished lifestyle or product content would be working against the stated point of view.
What would change this
A different engagement could work if the brand wants an analyst segment or written market read on why fashion and beauty are growing on Whatnot, rather than a creator-led selling format. That use matches the reviewed work. Confirm commercial terms and audience evidence before booking.
A worked campaign example
Angle
A research-backed webinar or article on what makes an AI shopping agent choose one product over another, using the agent-behaviour findings as the spine and ChatGPT's retail partnerships as marketplace context. The format is a 30-to-40-minute analyst session plus a written companion piece, with the creator as the named researcher-commentator rather than a sponsor presenter.
Close
End with the live-commerce comparison. Use the Whatnot evidence — roughly $8 billion in live ecommerce sales, growth above 100%, a claim of more than half the US market in its categories, about 95 minutes a day of user time, and fashion and beauty as the fastest-growing categories — to show another place where shopping behaviour is moving faster than most brands' plans. The takeaway is a short list of signals and listings to review, not a prediction about AI replacing stores.
Middle
Move from agent behaviour to seller behaviour. Walk through which listing signals appear to influence selection — price, rating, badges, and description wording — and which work in reverse, using the sponsored-label result as the clearest example. Then widen to the marketplace question: the second post argues that Etsy, Shopify, and Walmart already give ChatGPT broad inventory coverage, leaving gaps in electronics, fast fashion, furniture, and luxury. Its practical conclusion is that selling on Walmart or hosting on Shopify is the near-term route for brands that want to be reachable inside AI shopping surfaces.
Opening
Start with the finding, not the trend. The agents tested picked cheaper and higher-rated products, responded to "Overall Pick" and "Best Seller" labels, disagreed with each other, but consistently avoided anything tagged "Sponsored" — including sponsored results ranked first. The prompt used in the study said only to use best judgment, so the avoidance came from the model, not the instruction.
What to confirm before booking
Confirm commercial terms, as the posts do not specify a sponsor policy or rate card. Ask for reach or engagement on comparable analysis and audience demographic data to assess buyer-role fit. Confirm whether the creator will present live or record, since the posts show no video or event work. Secure permission to reproduce the cited research findings in the companion piece.
Why this works
Each segment rests on a source: the agent-behaviour study, the partnership rollout, and the platform sales figures. The session plays to the creator's strength — reading a source closely and turning it into a seller decision — rather than formats the posts do not demonstrate, such as live shopping or consumer product content.
Choose Sharon if the webinar or article should focus on how AI is changing the shopping journey, especially agent-led discovery, checkout and payments.
Read more
Her posts introduce agent-payment protocols and describe a shift from marketplace listings and social commerce toward agentic commerce, giving a campaign on that shift a broader consumer-journey frame.
Juozas Kaziukėnas’s posts survey the emerging landscape but do not offer the same combination of research into agent product choices, marketplace partnership analysis and Whatnot sales figures. A campaign centered on tested buying behavior and marketplace decisions would need additional evidence and examples.
Choose Haider if the buyer wants a practical demonstration of AI-assisted ecommerce research.
Read more
One post offers a reusable prompt and setup steps for comparing Amazon listings, reviews, brand stores and product gaps across browser tabs. That could make the webinar a hands-on seller workflow rather than a market-analysis session.
Juozas Kaziukėnas’s posts cover Amazon research methods and commerce protocols, but not the close analysis of AI-agent product selection or Whatnot market data that anchors a campaign on those topics. The protocol post also makes forward-looking claims, so verify them independently before relying on them.
Choose Laurie if the buyer’s main concern is marketplace operations and brand control, rather than how agents choose products or how live commerce is growing.
Read more
Her posts give concrete examples of unauthorized sellers, pricing violations, Buy Box loss, inventory diversion, listing changes and the risks of managing multiple sales channels separately. That could support a practical session on marketplace visibility and brand protection.
Laurie’s posts address control of listings, sellers and inventory—not AI-agent buying behavior or Whatnot’s growth. Laurie is the more relevant choice for an operations-led session, but the buyer would need another source or speaker to analyze agent behavior and live commerce.
