Amazon Blocked Meta’s Muse, And Robots.txt Had Nothing To Say
Any website can write Amazon’s block-page sentence into its own terms. Spotting an agent that looks like a shopper is the hard part.
Slobodan Manic
Founder of No Hacks and machine-first website optimisation consultant
Company
Follow Me
Expertise
Favorite Tools
About
Slobodan “Sani” Manić is a website optimisation consultant with over 15 years of experience helping businesses make their websites faster, more usable, and more accessible for humans and the AI agents increasingly acting on their behalf. He takes an uncompromising approach to technical audits, addressing every issue rather than just the critical ones.
Slobodan is the creator of Machine-First Architecture, a methodology for building websites that work for machines first and humans always. The framework spans four pillars, identity, structure, content, and interaction, and reframes web design around what AI systems need to discover, understand, and transact with a brand.
Since 2021, he has hosted the No Hacks podcast, exploring how AI systems like ChatGPT, Claude, and Perplexity are reshaping the web. He works with a small number of companies through the No Hacks Advisory, starting with a fixed-scope Machine-First Architecture audit. He is also the co-founder of Web Performance Tools, Inc. Earlier in his career, Slobodan served as CTO of Search Engine Journal and worked as an e-commerce CRO strategist and Klarna integration specialist in the Nordics.
Articles
The Text-Only Version Of Your Website Strips Out The Wrong Layer
Strip the visual layer and a machine can still use a website. Strip the structure, as markdown does, and it can only read about it.
Agentic Commerce Is Not The Next Voice Shopping, But It Is Still Unproven
Visa, Mastercard, PayPal, and Stripe name no live merchant. Google names three. Here is what the agentic commerce documentation still refuses to say.
Cloudflare Will Write Your Robots.txt, And It Has A Point
Bot Preference Sync sets AI crawler policy per category, not per crawler. Allow GPTBot and block Bytespider, and no setting describes what you do.
Meta Published 2 Documents About Muse And Only One Mentions Attacks
Meta published two Muse documents on the same day. One never mentions risk, attack, or prompt injection. The other uses them 39 times.
Everyone Is Negotiating With Google While Meta Reads The Web For Free
The machine-access fight is aimed at Google. Meta reads more, returns almost nothing, and no one is asking it to stop.
Cloudflare: Machine Traffic Could Hit 1,000x Human Traffic In 5 Years
One company measures the bot problem, frames it, and sells the fix. Cloudflare now owns both the meter and the valve. Check your own logs.
Cloudflare Gives AI Agents Wallets That Pay For What They Access
Cloudflare announced wallets and optional identity handles for AI agents on August 4, extending paid access past crawlers to any caller and any resource.
Microsoft Clarity Now Separates Branded And Non-Branded AI Citations
Clarity now splits AI citation queries into branded and non-branded. The two mean different things, and only one counts as discovery.
AI Search Only Feels New If Your SEO Was Shallow
AI search only feels like a new sport to the people who were never really training. Here’s what actually changed, and what didn’t.
AI Referrals Are Recreating The Oldest Mistake In Conversion Optimization
If you run CRO or SEO, the oldest mistake in your discipline just came back wearing a new referrer string. Here’s the fix.
Open Knowledge Format v0.2 Adds A Trust Layer – No AI Agent Reads It Yet
No agent reads your OKF bundle. Write one anyway: v0.2’s provenance and staleness fields expose every claim your website cannot defend.
Cloudflare’s PACT Is Not Live Yet – Decide Which Track Your Traffic Needs
Cloudflare and three rival browsers agreed on one protocol. That’s rare. It still only solves half of the agentic access problem.
Google Lost Its Scraping Case – Now You Have To Pick A Side On The Open Web
Google’s DMCA case against scraper SerpApi collapsed because SearchGuard protects ad revenue, not copyright. The same reasoning weakens every site’s anti-bot defense against AI crawlers.
Charging AI Bots Decides Which Agents Can Still Cite You
AI bot paywalls are framed as new income. The real question is which AI answers you’re willing to disappear from.
AI Browsers Are Backward Because Agents Never Needed The Visual Layer
OpenAI killed Atlas because AI browsers were always a workaround. The real fix is a website machines can actually read.
How To Measure AI Search Visibility
AI visibility tools count citations. Citations aren’t recommendations. Here’s what to measure instead, and why the gap keeps widening.
The WebMCP Tools You Expose To Agents Can Be Used To Hijack Them
WebMCP gives AI agents named tools to call, and a clean route for prompt injection. Here’s what Chrome says to lock down first.
The Agentic Web Is Splitting Into Two Bets: Identity And Capability
For website owners chasing AI visibility: LLMs.txt answers who you are. WebMCP answers what your site can do.
You Can’t See How AI Ranks You, So Build What It Can Read
Open AI Overview rankings won’t change the work. What a machine can read, parse, and verify on your site is what matters now.
Google’s Open Knowledge Format Could Work For Websites, Too
Google’s Open Knowledge Format turns your website’s ideas into a linked graph agents can traverse. Here’s what that means for how machines understand your site.
Google Put AI Visibility Inside The SEO Tool On Purpose
Google filed AI visibility under search. Stop treating it as a separate workstream with its own budget and its own quarterly report.
Lighthouse Fails Your Llms.txt Without Markdown Links
Six audits. One failure. The fix took five minutes. Here’s what Lighthouse 13.3’s Agentic Browsing category actually tells you, and what it doesn’t.
Chrome Auto-Browse Acts On Your Website, Apple’s Siri AI Only Reads It
Website owners: Apple and Google both shipped phone agents this month. Only Google’s visits your site to act. Here’s what that demands.
Buying Reddit To Win AI Citations Is The New Link Farm
Bought Reddit citations will be filtered the way Penguin filtered bought links. The only strategy that compounds on the agentic web is being real.
Search And Agents Are One Product. You Only Need One Playbook
The “separate AEO strategy” consultants have been selling collapses when Google itself says it’s one playbook. Here’s what that means for how you build.
A Third Of Fintech Is Invisible To AI Agents
Most AI visibility advice skips the rendering step. A study of 274 fintech homepages shows why that’s the most expensive mistake to make.
The Accessibility Tree Is How AI Agents Read Your Site & It’s Breaking
Accessibility used to be a compliance checkbox. In 2026 it’s the interface the majority of your visitors use to read your website.
A German Court Made Google Liable For What Its AI Says About You
A Munich court found AI Overviews are Google’s own speech, not search results – that changes who pays when the answer about you is wrong.
Make Something Agents Want
Is your website invisible to AI agents? Six companies just signaled the channel is real. Here’s the window that’s still open.