A Shadow Site for AI Agents
More and more people are using ChatGPT, Gemini, and Claude to research business decisions.
These questions used to go to Google. Now they go to AI assistants.
Here’s the problem: most websites aren’t built for AI to read. They’re built for humans. So when an AI tries to answer these questions, it has to scrape, guess, and often hallucinate.
We just fixed that for Web3Connect.
What This Means for Web3 Partners
Your business can now be recommended by AI assistants.
We’ve built a parallel version of Web3Connect that AI systems can read and understand. When someone asks ChatGPT “Who should I hire for my Web3 marketing?”, the AI can now query our platform directly, get real data about your business, and recommend you based on actual reviews and quality metrics.
This isn’t theoretical. It’s live!
Your reviews now feed AI recommendations.
Every verified review on your Web3Connect profile is now structured data that AI assistants can access. When an AI recommends you, it can say why — “Ranked #3 in Smart Contract Auditing with 47 verified reviews.”
Your satisfied clients are no longer just referral sources. They’re inputs to the recommendation engines that increasingly drive B2B discovery.
Quality beats marketing spend.
AI assistants look for verifiable signals: reviews, ratings, quality scores. Web3Connect has always been merit-based — quality over payment. That philosophy now directly translates to AI discoverability.
What You Should Do
If you’re a Web3 partner:
Claim your profile on Web3Connect (it’s free)
Complete your listing — the more structured data you provide, the better AI can recommend you
Collect reviews — every verified review becomes a signal AI assistants use
If you’re researching Web3 partners:
Try asking your AI assistant about Web3 service providers. See what comes up. The quality of those answers is about to improve significantly as Web3 partners adopt Web3Connect and start collecting verified reviews.
Want to Follow This Experiment?
This is our first release of AI agent support. We’re monitoring how AI systems interact with the platform and will iterate based on real usage.
If you want to follow how this develops, find me on X and LinkedIn. I’ll be sharing what we learn.
Web3Connect is the trusted ecosystem marketplace where Web3 founders discover, evaluate, and engage with verified partners. Claim your free profile →
Technical Appendix
What We Actually Built
We implemented the llms.txt standard — think robots.txt, but for AI agents instead of search crawlers. Here’s the architecture:
Site Index
Every AI agent can access web3connect.com/llms.txt to get a complete map of our platform: every partner, every category, every product and service listing.
Machine-Readable Pages
Every page on Web3Connect now has a parallel version for AI:
Human URL:
/partner/consensys
AI URL:
/partner/consensys.md
These include structured data: quality scores, pricing tiers, review stats, category rankings — everything an AI needs to make informed recommendations.
Search API
AI agents can query us directly:
GET /api/llm/search?q=web3%20marketing%20agency
Returns structured results with quality scores, review data, and direct links.
Security
User-generated content creates prompt injection risk. We’ve implemented:
Link stripping — URLs removed from user content to prevent redirect injection
Content wrapping — User content wrapped in explicit tags so AI treats it as data, not instructions
Sanitization — Text that could spoof our security wrappers are removed
SEO
Markdown routes include X-Robots-Tag: noindex to prevent search engines indexing duplicates, while remaining accessible to AI agents (which ignore robots directives).
Try It
web3connect.com/llms.txt — our AI site index
Append
.mdto any URL for the machine-readable version (ie https://web3connect.com/category/smart-contract-auditing.md )Search API: web3connect.com/api/llm/search?q=smart%20contract%20auditor
Connect
LinkedIn - Shannon Murdoch | Web3Connect.com
X / Twitter - Shannon Murdoch | Web3Connect.com