Your AI is about to go shopping

It's Thursday, and these are the only three AI stories worth your attention this week.

Let's get into it 👇

AI-powered shopping is coming for your wallet

Remember when buying stuff online meant going to a website? That's about to feel as outdated as a flip phone. Visa, Mastercard, and PayPal are building payment systems directly into AI chatbots, and it's going to change how we think about online shopping.

Here's the deal: You'll say something like "buy those Nike Air Force 1s in size 10" and the bot will check stock, process payment, and confirm shipping. No website required. You register your card once, and the bot handles the rest.

OpenAI isn't waiting around either. They're already adding direct product links to ChatGPT searches for fashion, electronics, and other stuff you might impulse buy at 1am.

Visa is calling this "agentic commerce" and comparing it to the Amazon or iPhone revolution. Normally I'd roll my eyes at that kind of hype, but ChatGPT went from zero to hundreds of millions of users in under three years. Regular e-commerce took decades to hit those numbers.

What this means for your business:

First, your product data needs to be spotless. AI engines will skip right over messy listings or incomplete descriptions.

Second, your ad strategy probably needs a total rethink. If bots start handling product discovery, those Google keywords you've been bidding on might not matter as much.

Last, whoever solves the trust problem wins. Visa anticipates real applications within a year, but the primary hurdle isn't technical capability. It's establishing trust. If consumers worry their AI shopping assistant might make unauthorized purchases, adoption will face significant resistance.

PRESENTED BY GUIDDE

🚀 Revolutionize Learning with AI-Powered Video Guides

Upgrade your organization training with engaging, interactive video content powered by Guidde.

Here’s what you’ll love about it:

1️⃣ Fast & Simple Creation: AI transforms text into video in moments.

2️⃣ Easily Editable: Update videos as fast as your processes evolve.

3️⃣ Language-Ready: Reach every learner with guides in their native tongue. Bring your training materials to life.

The best part? The browser extension is 100% free.

Check out Guidde

Unprecedented founder control in Thinking Machines' $2B raise

Remember when Sam Altman got briefly ousted from OpenAI last year? Mira Murati apparently took careful notes on that situation and created a governance structure that prevents similar scenarios.

Thinking Machines Lab just raised $2 billion in a Series A round led by Andreessen Horowitz, valuing this 18-month-old company at roughly $10 billion. But beyond the impressive numbers, the governance structure gives Murati an unprecedented level of control:

• Her single board vote outweighs ALL other directors combined

• Early researchers hold special shares with 100× voting rights and have given Murati their proxies

• She has the authority to appoint and remove directors at her discretion

For comparison:

• Meta: Zuckerberg has about 61% of votes with 10× shares

• Google: Brin and Page together have 52% with 10× shares

• Lyft: Founders control over 50% with 20× shares

• Thinking Machines: Murati has effectively total control with 100× shares

Corporate governance experts describe this setup as "highly unorthodox" in the world of venture-backed startups.

Why founders should care:

First, it demonstrates how significantly the power dynamics have shifted in the AI sector. The trend toward founder control has reached new heights.

Second, this creates an interesting test case for whether such concentrated authority produces better research outcomes. Will Murati's decisive control create the stability needed for breakthrough AI development?

Third, it signals that top AI talent now has unprecedented leverage in structuring new ventures. The traditional governance playbook is being rewritten.

The potential challenge? Public markets typically resist such concentrated control structures. And later-stage investors will have limited influence if strategic adjustments become necessary. But in the current AI landscape, Murati has established that elite AI leaders can negotiate governance structures that were previously unthinkable.

How one giant company avoided the enterprise AI money pit

Veolia has 220,000 employees and €44.7 billion in revenue. They've managed to roll out AI to 60,000+ users without their CFO jumping out a window. Here's their playbook:

1. Smart access tiers Basic stuff like text Q&A and email drafting? Sure, everyone can have that. Expensive stuff like image generation? Only for teams that actually need it.

2. DIY model marketplace They let employees pick their poison: Cohere, Mistral, Llama, or Claude. Everything runs through AWS Bedrock so billing and security stay centralized. Claude is what most people choose, but they use cheaper Stability AI for most image work.

3. Wholesale pricing strategy If they paid regular retail pricing ($25-30 per user per month), they'd be burning around $15 million every year. Their actual bill is "far below" that because of selective feature gates and volume discounts.

Steal this playbook for your company:

Only give expensive AI features to teams that actually make you money. Sorry, HR.

Let different departments pick the models they actually need for their work.

Run everything through one cloud provider so you can get better discounts and not fail security audits.

Start with a small pilot (Veolia began with just 2,000 users), watch what happens, then scale up.

Here's what's really interesting: smaller, task-specific models are making a comeback. They're cheaper and integrate more effectively with legacy systems. Matt Fitzpatrick from Invisible Technologies makes a compelling point: the real competitive advantage comes from rebuilding internal workflows around AI. This is where technology-native companies can establish significant advantages over traditional enterprises still adapting to digital transformation.

That’s a wrap for this week.