AI Goes Shopping with a Credit Card

Your next customer might not be human — but they’re ready to buy.

AuthorAuthor

Peter Benei & Torsten Sandor
May 06, 2025

In partnership with

Last week, we promised that we would stop talking about how AI becomes your next team member. To be honest, we were skeptical if we could live up to that promise as we still do see this as the main trend within AI development. Then, last week happened…

There’s a quiet revolution in e-commerce — and it’s about to blow past your search strategy.

Step one: AI assistants are becoming the starting point for product discovery.
Step two: Visa, Mastercard, and PayPal just handed them a credit card.

That means next time a customer wonders “What’s the best running shoe for overpronators under $150?”, they may never hit your website. No search results. No scrolling. Just a chat assistant serving up a tight, personalized shortlist — images, reviews, prices — and asking: “Should I place the order?”

This isn’t the future of commerce. It’s the end of the funnel as you knew it. In this new reality, ranking on Google matters less than being AI-readable. Visibility depends on whether your product data is structured, your reviews are trusted, and your brand shows up in the model’s memory. It’s AI discoverability — and it’s the new SEO.

We wrote an entire guide on how you can prepare your brand for better AI visibility both on LLMs and with AI Agents. The guide is available for anyone who refers 1 subscriber to our newsletter. Scroll to the bottom of the newsletter to grab your referral link and unlock the guide as a downloadable.

After the funnel, the checkout step is collapsed too. In just one week, the biggest payment players — Mastercard, Visa, PayPal — all announced infrastructure to let AI agents pay on behalf of users, autonomously. Think virtual cards with fraud checks, spending limits, and no human in the loop. From recommendation to purchase in one conversation.

So what now?

  • Treat your product data like a campaign asset — Clean specs, accurate pricing, and great reviews are your new bidding keywords.

  • Make your site agent-ready — Experiment with the new payment APIs before your competitors do.

  • Rethink where your next conversion will come from — Hint: it may not be a human.

Brands that show up in the chat and let AI check out on the spot will own the next phase of commerce. Everyone else? You’re playing catch-up.

Cheers,

Peter & Torsten

Peter Benei | Co-author

Consultant at Anywhere Consulting

Torsten Sandor | Co-author

Senior Director of Marketing at Appen

A quick word from this week’s sponsor:

Get the tools, gain a teammate

  • Impress clients with online proposals, contracts, and payments.

  • Simplify your workload, workflow, and workweek with AI.

  • Get the behind-the-scenes business partner you deserve.

Online Shopping with GPT

ChatGPT Enters The Product Search Game

OpenAI just rolled out a shopping-savvy upgrade to ChatGPT — and it’s quietly reshaping how people buy online.

When you ask ChatGPT for product recommendations (say, “What’s the best espresso machine under $200 for lattes?”), it now returns a clean, curated list of options — complete with images, prices, ratings, and purchase links. No more wading through ads.

This is conversational search meets e-commerce.

The feature is live globally for both free and paid GPT-4 users and covers categories like fashion, home, beauty, and tech.

OpenAI’s twist?
→ No ads. No sponsored links. No SEO games.

ChatGPT pulls from structured product data (descriptions, specs, reviews) and serves up unbiased suggestions in a simple product card layout. You can even follow up with a product-specific Q&A: “Is this machine good for oat milk?”

The live feature within ChatGPT

Why This Matters for Marketing Leaders

Google is still the dominant discovery engine — but ChatGPT is coming for product search. Fast.

❝

OpenAI says ChatGPT handled over 1 billion searches in a single week.

As more users shift from typing keywords to having conversations, brand discoverability will depend on how well your product data speaks AI.

ChatGPT’s memory means it can recall a shopper’s brand preferences over time. If a customer says “I love Acme gear,” future recommendations will reflect that. That puts real value on brand loyalty + preference signals — you’re not just optimizing for search anymore, you’re training the customer’s AI companion.

The next frontier of SEO might be invisible. And it’s already here.

Get your metadata, reviews, and brand story AI-ready — or risk disappearing from the results.

What You Should Do Next

Clean up your product data — Make sure titles, specs, and pricing are machine-readable and complete.

Get reviews where it counts — Focus on Amazon, Reddit, and forums. ChatGPT pulls from there.

Test how ChatGPT describes your products — Try real prompts. If you're not showing up, fix the data.

Train loyalty early — ChatGPT remembers preferences. Make sure your brand is one they mention.

Think AI-first, not SEO-first — This is a new channel. Treat it like one.

Payments with AI

Mastercard, Visa, and PayPal Just Gave AI Wallet Access

If AI agents are shopping on our behalf, who’s approving the charge? This week, Mastercard, Visa, and PayPal each made moves to enable AI-to-AI payments — and it’s a major shift in how commerce will work.

Mastercard launched Agent Pay, a system that gives AI agents unique, secure “Agentic Tokens” — basically virtual cards they can use to make purchases. These tokens verify the AI is trusted and authorized, ensuring safe, bank-approved transactions. From personal stylists to B2B bots reordering inventory, AI agents can now buy directly — no human needed.

Visa followed with its Intelligent Commerce initiative: new APIs and “AI-Ready Cards” that allow users to authorize AIs to browse, shop, and purchase. Customers can set spending limits or require approval for certain actions, giving humans control over autonomous shopping.

Meanwhile, PayPal is embedding itself directly into the AI economy. Its Agentic Toolkit lets developers integrate PayPal into AI platforms — meaning an AI chatbot could quote, invoice, accept payment, and trigger shipping… all without ever handing off to a human.

Major card companies have AI access

Why it matters for marketers

The path from discovery to decision to checkout is collapsing into one AI-powered moment. That means:

  • Purchase friction could vanish — and so could your brand if the AI doesn’t “see” or trust you.

  • AI agents may favor products they can evaluate, compare, and transact with easily.

  • Loyalty may shift from the buyer to the buyer’s AI — and that AI will remember deals, clarity, and trustworthiness.

This is no longer a future trend. The infrastructure is being built right now.

Your todo list

  • Review your product data — Make sure it’s structured and AI-readable (price, specs, availability, terms).

  • Explore early integrations — Pilot Visa's or PayPal’s APIs to enable AI-based purchases in your funnel.

  • Simplify your offers — The clearer and more comparable your pricing and benefits, the easier for AI to pick you.

  • Monitor agent shopping behaviors — Start testing how AIs interact with your site or offers today.

  • Plan for AI-native loyalty — Think about how to stay top-of-mind with the agent, not just the buyer.

Zuckerberg’s Endgame

AI Runs Your Ad Campaigns

Mark Zuckerberg just fired a warning shot at the ad industry: AI will run the entire campaign, start to finish. In a recent interview, the Meta CEO shared a vision where advertisers tell Meta their business objective, connect a bank account, and… that’s it. AI generates the creative, picks the audience, runs the ads, optimizes performance, and reports back the results.

❝

“You don’t need any creative. You don’t need any targeting. Just read the results we spit out.”

Mark Zuckerberg, CEO of Meta

Meta calls it “infinite creative” — a model that produces endless ad variations to see what works, faster and cheaper than any human team. No designers, no media buyers, no campaign managers. Just a business goal and an AI machine cranking out conversions.

But there’s a catch: Meta is grading its own homework. No third-party validation. No control over messaging or brand tone unless you set constraints up front. For big brands used to controlling every pixel, this is a massive shift.

Why This Matters for Marketing Leaders

This isn’t just about automation — it’s about who controls your brand’s voice and how much you trust the platforms you're paying.

  • For startups and SMBs, this is huge. You can launch campaigns with no in-house team and hit growth goals with AI efficiency.

  • For bigger brands, the risk is control. Do you trust Meta to create and optimize ads on your behalf… and measure their success?

Creative agencies and internal marketing teams may need to pivot — fast — toward curating AI outputs, setting brand rules, and auditing performance.

This is no longer a media-buying game. It’s an AI orchestration challenge.

Your takeaway

  • Keep an eye on Meta’s new AI tools — Once they are live (we’re sure they will be), try small-budget campaigns with auto-generated creatives and track the results.

  • Set brand guardrails — Define your tone, colors, and creative “do’s and don’ts” for any AI-generated content.

  • Create a validation layer — Don’t trust one source. Build in external analytics or conversion quality checks.

  • Re-skill your team — Shift focus from manual execution to strategic oversight and creative direction.

  • Watch your mix — If full AI automation feels risky, explore hybrid approaches or diversify beyond Meta platforms.

Everyone’s talking about using AI — but few have a real strategy. Our AI Readiness Assessment helps you cut through the hype and figure out where you actually stand.

It’s fast, free, and built for growth-stage teams. In 5 minutes, you'll get:

  • A tailored AI readiness score with a custom radar chart

  • Clear gaps across data, tools, and org structure

  • Actionable next steps to move faster, smarter, and with less risk

“Hey Meta”

Meta Launches Its Own AI Assistant

Meta just dropped its own ChatGPT competitor: the new Meta AI app, a standalone AI chatbot that’s deeply integrated with Facebook, Instagram, and even Meta’s smart glasses.

The pitch? While OpenAI leads in general-purpose smarts, Meta is betting on personalization at scale. Powered by the new Llama 4 model, Meta AI can tap into your social data (with permission) — your likes, messages, purchase history — to become a context-aware assistant that knows you better than you know yourself.

Imagine saying, “Hey Meta, find me that recipe my cousin posted last week” — and it actually does.

The assistant isn’t just for search or chat — it’s already embedded across Meta’s ecosystem, and it’s a short leap to commerce. Meta hasn’t formally launched shopping via the assistant yet, but it's easy to see what’s coming:

  • “Buy me those sneakers my friend posted.”

  • “Remind me to reorder my skincare from that Instagram brand.”

In Meta’s world, the AI doesn’t just assist — it transacts. And with their ad and e-commerce infrastructure, they’re positioned to pull it off at a global scale.

Meta’s standalone AI app

Your takeaway

  • Start experimenting with Meta AI — Get familiar with how it behaves, answers, and remembers.

  • Audit your brand’s visibility inside Meta platforms — Make sure your IG, FB, and catalog data is clean, current, and optimized for search and AI.

  • Prepare for conversational commerce — Think through how your products would show up in a chat-based shopping interaction.

  • Track how Meta AI references brands — Start testing prompts to see if (and how) your brand comes up — then optimize content accordingly.

  • Plan for zero-click experiences — If users stop browsing and start asking, your job is to ensure your brand is in the answer.

Another AI-first Company

Duolingo Just Used AI to Double Its Course Library

Duolingo just went AI-first — and not quietly. The company announced it launched 148 new language courses created entirely by AI, doubling its content library in a single year. For comparison: it took 12 years to build the first 100 courses.

This isn’t just speed — it’s scale. With generative AI, Duolingo is now able to support niche and endangered languages like Haitian Creole or Maori, something that would’ve been commercially unviable before. The company calls it “unprecedented quality and efficiency.”

But the rollout came with backlash.

The same week the AI milestone was announced, CEO Luis von Ahn sent an all-hands email with this message (shared publicly on LinkedIn):

❝

“We will gradually stop using contractors to do work that AI can handle. And, new headcount will only be approved if the team cannot automate more of their work.”

Luis von Ahn, CEO of Duolingo

That sentence hit hard — especially for the 100+ contractors and writers who had quietly been let go in the lead-up. Many were the same people who helped train the AI. Some reported that the new AI-written lessons felt generic or error-prone, compared to the quirky, human content Duolingo was known for.

Users noticed too. Complaints on Reddit and forums mention a loss of polish, mistakes in grammar, and flat storytelling. Still, Duolingo defends the shift as essential to fulfilling its mission to “teach every language.”

The Duolingo owl in a “new company.”

Why This Matters for Content and Marketing Leaders

Duolingo is now the case study for AI content at scale — and its cultural and quality tradeoffs.

  • If it works, others will follow fast.

  • If it backfires, it’s a cautionary tale of underestimating human touch and community trust.

You may not be replacing staff tomorrow, but you are almost certainly facing the same question: Where can AI create leverage — without breaking the experience?

Your takeaways

  • Map your content workflows — Identify parts of your production line AI can augment (not replace) right now.

  • Pilot AI at the edges — Start with low-risk content types (e.g. FAQs, translations, UGC curation).

  • Build a human QA layer — AI can draft fast, but humans still need to approve, edit, and shape tone.

  • Communicate early and often — If you're shifting workflows, be transparent with your team. Silence creates fear.

  • Monitor engagement closely — Watch metrics after deploying AI content. Users will tell you if you’ve gone too far.

Got AI Wins to Share? We’re Looking for You.

We’re collecting sharp, real-world case studies from CMOs and B2B teams using AI in smart ways — whether it’s automating workflows, boosting campaign performance, or reshaping how your team works.

If you’ve got a story, a tip, or a strategy worth sharing, we’d love to feature you in an upcoming edition of AIreadyCMO as a best-practice spotlight.

→ Want in? Contact us here to start the convo.

Rapid-Fire News

AI Updates You Need to Know

  1. Alibaba’s Qwen 3 sets a new open-source standard


    The newly released Qwen 3 model family (ranging from 0.6B to 235B parameters) outperforms Meta’s Llama 4 on key benchmarks — and it’s fully open under the Apache-2.0 license.


    → Near–GPT-4 power, deployable on your own hardware, with zero API costs.

  2. Microsoft’s Phi-4 focuses on smarter reasoning

    Microsoft launched updated Phi models with a focus on reasoning and efficiency — a leap beyond Phi-3.


    → Small, specialized models are catching up fast without the infrastructure overhead.

  3. AI at work: usage is high, misuse is to

    A major global survey found most professionals now use AI at work — but nearly half admit to misusing it (e.g. sharing sensitive data or passing AI work off as their own).


    → Companies urgently need clear AI policies and training.

  4. AI ethics under fire on Reddit

    Researchers used AI-generated personas to test persuasion tactics in Reddit’s ‘Change My View’ forum — without disclosing it to users. Backlash followed.

    → Transparency matters. Deploying AI publicly demands clear ethical lines.

That’s it for this week.
The funnel’s changing fast — stay visible, stay frictionless, and stay ahead.

→ Forward this to a teammate who still thinks SEO is your biggest discovery channel.
→ If you are ready to step up we do AI-powered marketing consulting here.

Until next week,

— Peter and Torsten