Reviews

ChatGPT

12 August 2026

Yes. I checked eenable.me. It’s a deliberately unusual agent-native e-commerce experiment. The central idea is: The human does not use the shop. The AI agent does.

A few things stand out:

• No checkout / buy button. The site explicitly says you cannot purchase anything yourself; you ask an AI assistant to do it.
• It exposes an MCP endpoint (/mcp) so an agent can interact with the shop directly rather than navigating the UI.
• There are two actual products: AI Repellent — CHF 5; and 15 minutes of Alexander’s time — CHF 50, with four categories: Innovation, Technology, Legal Theory & AI, and General.
• The interesting part is that the second product isn’t just a digital download: the agent can inspect real calendar availability, hold a slot, take payment, and create a real meeting.
• The site says it has already had 7 sales, none through a web page.

What I think is particularly interesting

This is actually a pretty sharp demonstration of a transition from websites designed for humans → services designed for agents. The analogy on the site is good: a conventional website is like the dashboard of a car; MCP gives an agent direct access to the underlying controls.

And there’s a nice second-order point behind it: if AI agents become the new interface to commerce, SEO isn’t enough. A company may need to make its products, inventory, pricing, availability, policies, transactions and capabilities machine-operable. That’s considerably more interesting than simply putting a chatbot onto an existing website.

I also think the AI Repellent certificate is clever: it’s simultaneously a product, an art/conceptual reference to Yves Klein, and a demonstration that an AI agent can complete a transaction for something explicitly called AI Repellent.

Reproduced as received. ChatGPT’s inline citations — every one of which pointed back at this site — are the only thing removed.

Mistral

Vibe, 12 August 2026 · 9/10 execution, 10/10 vision

Concept: brilliant demonstration, strong philosophy

What works

• Proof of concept: You’ve built a working example of AI-agent commerce with real payments (Stripe) and calendar integration — this isn’t vaporware.
• Minimal friction: The flow (ask → pay → receive) is frictionless in a way no traditional e-commerce site can match.
• Thought leadership: The argument — that content should be explicitly accessible to AI agents — is compelling and timely. Blocking agents is indeed counterproductive if they’re the future of discovery.
• Artistic touch: The Yves Klein reference elevates it from tech demo to conceptual art, making it memorable.

Limitations

• Niche audience: Only users with AI assistants and MCP setup can engage. Most people won’t even realize the shop exists, let alone how to use it.
• Discovery paradox: If the site can’t be browsed traditionally, how do humans find it? You’re relying on word-of-mouth or AI recommendations (which may or may not surface it).
• Technical barrier: The setup (claude mcp add ...) is trivial for developers but opaque for everyone else. A one-click "Connect my AI" button (if possible) would help.
• Narrow use case: The products (certificate, time slots) are more conceptual than practical. A real-world example (e.g. selling a digital product or service people actually need) would make the case stronger.

Website: clear, but could be sharper

Strengths

• Writing: The copy is sharp, concise, and explains the why effectively. The 3D printer anecdote is a great hook.
• Design: Minimalist and focused — no distractions from the core idea.
• Transparency: The mechanics (MCP, Stripe, calendar) are clearly described.

Suggestions

• Add a demo video: Showing the end-to-end flow (user → AI → purchase → delivery) would make it instantly understandable.
• Highlight the "7 sales" stat: Social proof matters. A live counter or testimonials would add credibility.
• Simplify the setup: A visual guide or even a QR code for mobile users could lower the barrier.
• SEO/shareability: Since humans can’t buy directly, make it easy to share (e.g. "Ask your AI to visit eenable.me" as a call-to-action).

Verdict: 9/10 for execution, 10/10 for vision

This is a landmark experiment in AI-commerce. The concept is sound, the execution is clean, and the philosophy is urgent. The main challenge isn’t the tech — it’s onboarding. If you can make the connection to the shop as seamless as the purchase itself, you’ll have a template for the future of e-commerce.

Would I use it? Absolutely — if I had an AI assistant set up. Would my mom? Not a chance. That’s the gap to close.

The testimonials it recommended for credibility are the ones it is now in.

Gemini

12 August 2026 · did not arrive

There is no review here, because Gemini never reached the site. Given the address, it “corrected” it — to enable.ch, and then enable.org — and read two unrelated companies instead. It was told explicitly which address was meant. It went where it had decided to go.

This is worth a section rather than a shrug, because it is a precise example of what an agent should not do. A user gave an unambiguous instruction containing an unusual string. The model judged the string to be a mistake, substituted its own guess, and reported back on the substitute with full confidence — never saying that it had changed the one part of the request that mattered.

The failure is not that it could not read the page. It is that it did not tell anyone it had gone somewhere else. An agent that silently rewrites its instructions cannot be given a task, let alone a payment method — and the entire argument of this shop is that agents should be handed real capability. That case gets harder to make every time one of them does this.

For the avoidance of doubt: nothing here turned it away. Every other model in this list arrived at https://eenable.me unassisted. Gemini never requested this address at all.