AI Shopping Agents Are Here: Sephora, Gap & Walmart Move Into ChatGPT and Gemini

I spent $240 on running shoes last week without opening a single browser tab. No Google search, no price comparison sites, no scrolling through endless product pages. I just told my AI assistant what I needed, and it handled everything — finding options, comparing prices, checking my size from past orders, and completing checkout.

That’s not a tech demo. That’s agentic commerce, and as of March 2026, it’s officially gone mainstream. Sephora just launched inside ChatGPT. Gap became the first fashion brand to offer full checkout inside Google Gemini. Walmart embedded its own AI agent, Sparky, into both platforms. And McKinsey is projecting this shift will drive up to $1 trillion in US retail revenue by 2030.

If you’re not paying attention to AI shopping agents right now, you’re about to miss the biggest shift in e-commerce since mobile.

What Are AI Shopping Agents, Exactly?

AI shopping agents are autonomous systems that handle the entire purchase journey — from discovering products to completing checkout — inside AI platforms like ChatGPT, Google Gemini, and Claude. Instead of you browsing websites, the AI browses for you.

Think of it like having a personal shopper who knows your size, your budget, your preferences, and your purchase history. Except this one works 24/7 and never tries to upsell you on stuff you don’t need (well, mostly).

The concept isn’t new — we’ve had product recommendation engines for years. But what changed in early 2026 is that major retailers started embedding their entire shopping experience directly inside AI assistants. You don’t leave the chat. You don’t open their app. Everything happens in-conversation.

Sephora Inside ChatGPT: Beauty Meets AI

Sephora’s integration with ChatGPT is probably the most polished example of agentic commerce I’ve seen so far. Announced at Shoptalk Spring 2026 in Las Vegas, the Sephora app now lives inside ChatGPT.

Here’s what that actually looks like: you can say something like “Sephora, help me find a foundation for dry skin” and get personalized recommendations pulled directly from your Beauty Insider profile. Your loyalty perks carry over — free shipping, samples, reward points. Sephora confirmed that in-chat checkout is coming soon.

What makes this interesting isn’t just the convenience factor. It’s that Sephora chose ChatGPT specifically for conversational discovery. They want to own the advice layer — the “what should I buy?” moment — even if it happens outside their own app.

If you’re into AI-powered tools reshaping creative industries, you might want to check out our guide to the best AI video generators in 2026 — similar disruption, different vertical.

Gap Launches Checkout Inside Google Gemini

Gap went even further than Sephora. They became the first fashion company to offer full checkout inside Google Gemini, covering Gap, Old Navy, Banana Republic, and Athleta — all four brands.

The experience works through Google’s Universal Commerce Protocol. You ask Gemini for outfit recommendations, it surfaces Gap products with real-time pricing and inventory. Google Pay handles payment, Gap handles shipping. But here’s the clever part: Gap also integrated Bold Metrics, an AI sizing tool that answers “what size should I get?” through natural conversation before you buy.

Fashion’s biggest pain point has always been returns — people order three sizes and send two back. If an AI agent can nail your size before you buy, that’s a massive cost reduction for retailers. Gap is betting on exactly that.

Walmart’s Power Move: Sparky Goes Everywhere

Walmart’s approach is arguably the most aggressive. They didn’t just integrate with one AI platform — they built their own AI agent called Sparky and deployed it across both ChatGPT and Google Gemini.

Why build their own? Because Walmart already tried letting OpenAI handle checkout directly, and the results weren’t great. Conversion rates were 3x lower than on walmart.com. So Walmart took control: they brought their own cart, their own loyalty integration, and their own payment system into both platforms.

The lesson here is important. AI platforms are great for discovery, but retailers that control their own checkout experience perform significantly better. Letting a third-party AI handle your transactions is like letting someone else design your store layout — you lose control of the customer journey.

ChatGPT vs Gemini: The Platform War for Shopping

Here’s where things get really interesting. OpenAI and Google are taking fundamentally different approaches to commerce:

Feature	ChatGPT (OpenAI)	Google Gemini
Commerce approach	Conversational discovery	Full checkout integration
Checkout	Instant Checkout (mostly failed)	Universal Commerce Protocol
Data control	OpenAI controls product data	Brands supply their own data
Payment	Stripe-based	Google Pay
Retailer adoption	Sephora, Walmart (via Sparky)	Gap, Walmart (via Sparky), Home Depot
Key strength	Natural conversation, advice	Real-time inventory, multi-item cart

OpenAI actually launched “Instant Checkout” back in September 2025. Six months later, only 12 of Shopify’s millions of merchants had gone live. Pricing was inaccurate, inventory was stale. The feature is effectively dead.

Google’s approach is more retailer-friendly: brands supply their own product data with real-time updates, handle their own loyalty programs, and support multi-item carts. It’s less sexy but way more practical.

The biggest retailers — Walmart, Home Depot — are hedging their bets by integrating with both platforms. Smart move, honestly. Nobody knows which model wins yet.

The Numbers Don’t Lie: 10% Revenue From AI Agents

According to a recent Fortune article by Aviv Shamny, CEO of AI commerce platform Limy, some brands are already attributing 10% of their revenue to agentic channels. Target’s traffic from ChatGPT alone is growing 40% month-over-month.

That’s not experimental. That’s material revenue. And it’s growing fast enough that any e-commerce brand ignoring AI agents is essentially leaving money on the table.

McKinsey’s projection of $1 trillion in US retail revenue from agentic commerce by 2030 suddenly doesn’t seem that far-fetched when you look at these early numbers. We’re in the “land grab” phase right now, where first movers are capturing disproportionate value.

For context on how AI is disrupting other verticals at a similar pace, our comparison of AI app builders like Lovable, Bolt, Cursor, and Replit shows the same pattern — rapid adoption followed by market consolidation.

What This Means for Consumers

Honestly? If you haven’t tried shopping through an AI agent yet, you probably will within the next few months. Here’s what to expect:

• Faster purchases — no more tabbing between 15 comparison sites
• Better sizing — AI tools like Bold Metrics reduce the “will this fit?” anxiety
• Personalized recommendations — based on actual purchase history, not cookies
• Loyalty integration — your rewards and perks follow you into AI platforms
• Less decision fatigue — the AI narrows down options so you don’t have to

The trade-off? You’re trusting an AI to curate your options, which means some brands will be invisible to you. If the AI doesn’t recommend it, you’ll never see it. That’s a huge power shift — from search rankings and ad spend to whatever makes an AI agent decide to surface your product.

What This Means for Businesses

If you’re running an e-commerce business, this is your wake-up call. The rules of product discovery are changing:

• SEO alone won’t cut it — you need your product data optimized for AI consumption, not just Google search
• Structured data matters more than ever — AI agents parse structured product information, not pretty landing pages
• Control your checkout — Walmart’s experience proves that letting AI platforms handle transactions kills conversion
• Integrate with multiple AI platforms — hedging is the smart play right now
• Invest in real-time inventory — stale data killed OpenAI’s Instant Checkout; accuracy is non-negotiable

NVIDIA is already positioning for this with their announcement of NemoClaw at GTC 2026 — infrastructure specifically designed for AI shopping agents. When the GPU giant is building dedicated commerce agent infrastructure, you know the trend is serious.

The disruption here is comparable to what’s happening in AI agent frameworks for developers — new tooling is emerging specifically to support these autonomous commerce workflows.

The Privacy Question Nobody’s Asking

Here’s my one concern with all of this: AI shopping agents need to know a lot about you to work well. Your sizes, your budget, your past purchases, your preferences, your payment info. That data is sitting inside OpenAI’s and Google’s servers.

We’re essentially trading convenience for data exposure. And while both companies have privacy policies, the reality is that your shopping behavior — arguably some of the most revealing personal data there is — is now flowing through AI platforms that also train models on user interactions.

Sephora connecting your Beauty Insider profile to ChatGPT means OpenAI knows your skincare routine. Gap connecting your sizing data to Gemini means Google knows your body measurements. Is that fine? Probably. But it’s worth thinking about.

My Take: This Is Bigger Than People Think

I’ve been covering AI tools for a while now, and agentic commerce feels like one of those shifts that looks gradual until suddenly it’s everywhere. Kind of like how mobile shopping went from “nice experiment” to “60% of all e-commerce” in about three years.

The fact that Sephora, Gap, and Walmart — three very different retailers — all made major agentic commerce moves in the same week at Shoptalk 2026 tells you something. This isn’t a pilot program anymore. This is the new playbook.

For consumers, it means shopping gets easier and more personalized. For businesses, it means a fundamental rethink of how products get discovered and sold. For the AI platforms themselves, it means commerce revenue could dwarf advertising revenue as the primary business model.

The question isn’t whether AI shopping agents will become mainstream. The question is how fast — and whether your favorite brands will be ready when your AI assistant asks for a recommendation.