First published December 2024. Updated for 2026 with what actually happened.
When this article first ran in December 2024, the question of AI Christmas shopping was pure speculation. Could an artificial intelligence really browse, choose, and buy your gifts for you? Eighteen months later, we no longer have to guess. We can answer with evidence.
The short version: the agents arrived almost exactly on schedule — and then ran straight into the two problems this article warned about in 2024. The story of AI Christmas shopping in 2025 and 2026 is not a clean triumph. It is a far more interesting tale of a future that half-arrived.
In January 2025, on a livestream, a user held a handwritten grocery list up to a camera. An AI called Operator read it, opened Instacart, built the cart, and scheduled the delivery — no typing, no searching. The audience applauded. For a moment, fully automated AI Christmas shopping looked inevitable. What happened next is the real lesson.
The 2024 Prediction, and What Actually Happened
The original version of this article made a simple bet: that “agentic AI” — systems that don’t just answer questions but take multi-step actions in the real world — would soon move from demo to daily life, and that holiday shopping would be one of the first places it landed.
That bet was largely correct on timing and wrong on ease. The tools for AI Christmas shopping did appear in 2025. What nobody fully appreciated in 2024 was how hard the unglamorous parts — payment, trust, real-time product data, and context — would prove to be.
So this update does two things. It tells you what genuinely works now, in 2026, and it explains why the dream of telling an AI “handle Christmas” and walking away is still, for most people, a year or two off.
The Agents Arrived On Schedule
In January 2025, OpenAI released Operator, its first true agentic tool. Unlike a chatbot that only talks, Operator could click, scroll, and type its way through a website to complete a task — including shopping. It launched with retail partners including eBay, Etsy, Instacart, DoorDash, Uber, Priceline, and StubHub.
At the launch, OpenAI’s chief executive Sam Altman was candid that the tool still had a long way to go, but that the company wanted it in people’s hands. That honesty turned out to be well judged.
Anthropic had already shown the way three months earlier, in October 2024, with a “computer use” capability that let its Claude models operate a keyboard and mouse. Google followed with Project Mariner, an agent that worked inside the Chrome browser. The building blocks for AI Christmas shopping were suddenly real, and every major lab was racing to assemble them.
What an AI Shopping Agent Actually Does
Strip away the hype and a shopping agent performs a chain of distinct jobs. Understanding the chain is the easiest way to see both the promise and the weak links.
| Step | What the agent does |
|---|---|
| Understand | Reads your request and past preferences to infer what you actually want |
| Research | Searches across stores and marketplaces for matching products |
| Compare | Weighs price, reviews, shipping time, and availability |
| Decide | Recommends a choice and asks you to confirm |
| Buy | Completes payment and checkout |
| Manage | Tracks delivery and handles problems or returns |
In 2024, every step beyond “research” and “compare” was theoretical. By 2026, the “buy” step works — but only in narrow, carefully wired conditions. The reason is a piece of invisible plumbing that did not exist when this article was written.
The Plumbing That Made AI Christmas Shopping Real
An agent screen-scraping a shop is fragile. Crucial details — live inventory, real shipping costs, whether an item is even in stock — often are not visible on the page. So in 2025 the industry started building proper rails for machines to transact.
In September 2025, OpenAI and the payments company Stripe launched Instant Checkout, powered by an open standard they called the Agentic Commerce Protocol. It let ChatGPT users buy directly inside a conversation, starting with Etsy sellers and expanding toward Shopify merchants. Walmart joined as an anchor partner that October.
Google answered in January 2026 with the Universal Commerce Protocol, unveiled at the National Retail Federation conference with a coalition including Shopify, Etsy, Target, Wayfair, Walmart, Visa, Mastercard, and Stripe. Microsoft launched Copilot Checkout. The race to own AI Christmas shopping had turned into a standards war.
These protocols are the genuinely important development. They are the language that lets an AI agent, a shopper, and a merchant complete a purchase together — the foundation everything else now sits on. This is the same machine intelligence explored in our guide to how large language models work, now pointed squarely at your wallet.
The Big Players Racing to Own AI Christmas Shopping
The contest to dominate AI Christmas shopping now involves every major technology company, each with a different strategy and a great deal of money at stake.
OpenAI moved first with Operator and Instant Checkout, betting on buying inside the ChatGPT conversation that already draws around 700 million weekly users.
Google has folded shopping into Gemini and its AI search mode, leaning on decades of merchant relationships and its Universal Commerce Protocol. Many analysts see it as the front-runner precisely because it already knows both the products and the shoppers.
Amazon launched its own “Buy for Me” agent in 2025, alongside its Rufus shopping assistant, letting customers buy items from outside Amazon within the Amazon app. It also moved to defend its turf, suing the AI search company Perplexity to block its agent from operating inside Amazon’s marketplace — an early sign that the agentic future will be fought in courtrooms as well as labs.
Microsoft built Copilot Checkout with Shopify, PayPal, Stripe, and Etsy, while Anthropic, maker of Claude, has leaned toward helping merchants and small businesses rather than shopping for consumers directly.
Walmart has taken perhaps the canniest position of all, building its Sparky assistant to plug into ChatGPT, Gemini, and any other agent — betting that whoever controls the catalogue and fulfilment wins no matter which AI sits on top. That makes the retailer, not just the AI lab, a central player in AI Christmas shopping.
A Short Timeline: How AI Christmas Shopping Got Here
- Oct 2024 — Anthropic releases Computer Use; Claude can operate a browser.
- Dec 2024 — Google announces Project Mariner.
- Jan 2025 — OpenAI launches Operator with retail partners including Etsy, Instacart, and eBay.
- April 2025 — Amazon launches “Buy for Me”; Google publishes its agent-to-agent protocol.
- June 2025 — Anthropic’s Project Vend tests an autonomous shop for a month.
- Sept 2025 — OpenAI and Stripe launch Instant Checkout and the Agentic Commerce Protocol.
- Oct 2025 — Walmart joins as an anchor partner.
- Jan 2026 — Google and partners launch the Universal Commerce Protocol; Microsoft Copilot Checkout goes live.
- March 2026 — OpenAI retires native Instant Checkout, pivoting to retailer-run ChatGPT Apps.
The Stumble: Why Full Autonomy Is Still Hard
Here is the twist the 2024 hype missed. In March 2026, barely six months after launching it, OpenAI quietly retired Instant Checkout in its original form. Only around thirty Shopify merchants had fully integrated, and shoppers were not converting at the scale anyone expected.
The replacement model moved away from buying inside the chat and toward “ChatGPT Apps,” where retailers like Walmart, Target, Etsy, and Instacart run their own experiences and keep control of the transaction. The infrastructure was built. The merchants were ready. The users simply were not converting at scale.
Industry insiders were unusually frank about it. One widely shared assessment held that nobody in the field had shopping agents truly figured out — each player merely assumed a rival did. There were practical gaps too: even basic questions, such as how sales tax would be collected and remitted on in-chat purchases, remained unresolved well into 2026.
Trust, Hallucinations, and Context
The 2024 version of this article flagged two obstacles to useful shopping agents. Both have aged remarkably well.
The first is trust. Would you hand an AI your saved card and let it buy unsupervised? Hallucination — the tendency of AI to state confident falsehoods — remains a live risk. One widely cited study of AI-generated programming answers found that a majority contained errors. In shopping terms, one mistake means Uncle Joe receives Aunt Molly’s gift, or worse.
The second is context. Buying a good gift is not a logistics problem; it is a knowing-someone problem. A parent who has understood a child’s tastes for twenty years carries context no agent yet holds. The labs are closing that gap by gathering ever more personal data — which raises its own uncomfortable questions, the kind we examine in our piece on the high-stakes battle over AI training data.
The deeper an agent’s grasp of context becomes, the more it edges toward the general capability discussed in our explainer on artificial general intelligence — and the further it travels from a simple shopping tool.
Can You Trust an Agent With Your Money?
The single biggest barrier to AI Christmas shopping is payment trust, and the industry knows it. Rather than hand raw card numbers to an AI, the new systems lean on the established payment networks.
Visa and Mastercard have both launched agentic-payment frameworks, and providers like Stripe and PayPal sit between the agent and your bank, using tokenised credentials and spending limits rather than exposing your details to the model.
PayPal has extended buyer protection to AI-initiated purchases, and the open protocols are designed so the merchant — not the AI — remains responsible for fulfilment, returns, and support.
None of this makes an agent infallible. But it means a well-built AI doing your Christmas shopping is not freelancing with your bank details; it is passing a tightly controlled instruction through the same rails that already move trillions in card payments every year.
The Money Behind AI Christmas Shopping
If the consumer experience is still rough, the money behind it is deadly serious. The projections are enormous, and they explain why every major player is pushing so hard.
McKinsey has estimated that agentic commerce could redirect three to five trillion dollars of global retail spending by 2030, with around a trillion of that in the United States alone. Bain & Company projects that AI agents could handle 15 to 25 per cent of US e-commerce by the end of the decade.
Consumer behaviour is already shifting. Morgan Stanley has reported that roughly a fifth of Americans made a purchase via AI in a single recent month. Adobe found that traffic to retail sites from AI tools surged several-fold over the 2025 holiday season, and eMarketer expects AI platforms to drive around twenty billion dollars in retail spending in 2026 — close to four times the previous year.
The behavioural signal is just as striking. OpenAI has said ChatGPT alone now fields tens of millions of shopping-related queries a day, and one analysis found shoppers arriving from AI tools were markedly more likely to buy than those coming from a traditional search engine. AI Christmas shopping is no longer a niche experiment; for a growing share of people it is becoming the first stop on the way to the checkout.
This is why a quiet new discipline has appeared: making sure an AI agent can find and understand your products. It is the agentic successor to search-engine optimisation, and the same pattern-recognition that drives it powers the neural networks beneath every modern AI assistant.
So, Will AI Do Your Christmas Shopping in 2026?
The honest 2026 answer: AI will help you shop far more than it did in 2024 — and still will not do it for you while you sleep. The agents are real. The autonomy is not yet.
For discovery and ideas, AI Christmas shopping is already excellent. Ask a good model for “gifts for a ceramics lover under fifty pounds” and you will get a thoughtful, well-reasoned shortlist in seconds. For comparison and research, it saves real time.
For the actual purchase, you can now, in some cases, buy with a tap inside a conversation — but with your approval at each meaningful step, not hands-off. And for the dream of “handle Christmas entirely,” the technology is close on capability and still short on trust, context, and reliability.
The trajectory is clear even if the timeline is not. Within a few years, sharing a recipe may trigger the groceries to arrive, and a wedding on your calendar may prompt a quiet shortlist of dresses. As the original version of this article put it, agentic AI is evolving rapidly, and it holds the potential to reshape daily life like few technologies before it. Two years on, that judgement looks not just intact, but understated.
How to Use AI for Your Christmas Shopping Today
You do not need to wait for full autonomy to benefit. Used well, today’s tools already make AI Christmas shopping genuinely easier. A few practical habits work right now.
Start with discovery, not checkout. Describe the person — their age, interests, the last gift they loved, your budget — and ask for a shortlist with reasons. The more context you give, the sharper the suggestions.
Use AI to compare, then buy yourself. Ask it to weigh two or three options on price, reviews, and delivery time. Let it do the legwork; keep the final click for now.
Keep a human approval step. Where an assistant can complete a purchase, confirm the order, address, and total before it goes through. Treat the agent as a capable helper, not an unsupervised buyer.
Mind your data and permissions. Give a shopping agent the narrowest access it needs, prefer tools that route payments through trusted processors, and check what the assistant already knows about you.
Protect the joy. Hand the agent the dull logistics — predictable staples, restocks, price tracking — and keep the meaningful, personal gifts for yourself. The best use of AI Christmas shopping is to clear time for the parts that actually matter.
What Comes Next for AI Christmas Shopping
The direction of travel is set even if the speed is not. The commerce leads at the major AI companies say the tipping point is months, not years, away — though similar things were said in 2024, so a little caution is wise.
The likeliest near-term shape is not a single all-powerful agent but a layered one: a retailer’s assistant, such as Walmart’s Sparky, travelling into ChatGPT, Gemini, and other front ends through shared protocols. You will meet AI Christmas shopping less as one app and more as a capability woven through tools you already use.
Three things should improve fastest: real-time product data, so agents stop recommending out-of-stock items; payment trust, as tokenised and protected transactions mature; and personalisation, as assistants learn your tastes. When those three line up, the friction that still surrounds AI Christmas shopping will quietly fall away.
The 2024 prediction was right that this future was coming. The lesson of 2025 and 2026 is that it arrives in fits and starts, with the occasional embarrassing pullback — not in a single magical Christmas morning.
Frequently Asked Questions
Can AI do all my Christmas shopping for me right now?
Not fully and not unsupervised. As of 2026, AI is excellent for gift ideas, research, and comparison, and in some cases can complete a purchase with your approval at each step. Fully hands-off AI Christmas shopping — telling an agent to “handle it” and walking away — is not yet reliable for most people.
Which AI tools can actually buy things in 2026?
ChatGPT (through retailer apps and the Agentic Commerce Protocol), Google Gemini (via the Universal Commerce Protocol), Microsoft Copilot, and Perplexity can all complete real purchases in supported cases. Amazon offers its own Rufus assistant and “Buy for Me” agent. Capability varies by region and retailer.
Is it safe to let an AI agent use my card?
Only with strong safeguards and limited permissions. New protocols route payments through established providers like Stripe, Visa, and Mastercard rather than handing raw card details to the AI, and buyer-protection policies are being extended to AI-initiated purchases. Even so, keeping a human approval step is wise.
Why did OpenAI’s Instant Checkout get pulled back?
Despite heavy retailer interest, relatively few merchants had fully integrated and shoppers were not converting at scale. In March 2026 OpenAI shifted from buying natively inside the chat to “ChatGPT Apps,” where retailers run their own in-chat experiences. The underlying Agentic Commerce Protocol survives as infrastructure.
Will AI Christmas shopping reduce holiday stress?
For organisation, research, and shortlisting, almost certainly yes. But it may also strip away some of the personal joy of choosing a gift yourself — and an over-trusted agent that makes a mistake can create more stress than it saves. Used as an assistant rather than a replacement, it is genuinely helpful.
What is the Agentic Commerce Protocol?
An open standard co-developed by OpenAI and Stripe in 2025 that lets an AI agent, a shopper, and a merchant complete a purchase together. It passes order details to the merchant’s existing systems so the retailer keeps control of payment and fulfilment. Google’s rival Universal Commerce Protocol does a similar job, and these standards — not any single app — are the real foundation of AI Christmas shopping.
Is AI shopping available outside the United States?
Most of the in-chat checkout features launched in the US first, with merchants and regions expanding over time. Discovery, research, and recommendation features work almost everywhere, but the ability to complete a purchase directly through an AI still depends heavily on your country and the specific retailer.
Further Reading
Sources
- OpenAI — Instant Checkout and the Agentic Commerce Protocol
- Anthropic — Computer Use announcement
- Google DeepMind — Project Mariner
- CNBC — OpenAI’s agentic shopping pivot (2026)
- Fast Company — The race to build agentic commerce
- Deloitte — Global AI Predictions (2025)
Baryon. (2024, December 24). Will AI do your Christmas shopping in 2026? Current Capabilities and Future Possibilities. Web News For Us. https://webnewsforus.com/ai-christmas-shopping/
Baryon. “Will AI do your Christmas shopping in 2026? Current Capabilities and Future Possibilities.” Web News For Us, 24 December 2024, https://webnewsforus.com/ai-christmas-shopping/. Accessed 8 July 2026.
