Over 40% of search traffic is no longer human eyeballs typing keywords into Google. Autonomous AI agents powered by OpenAI Operator, Perplexity Pro, Anthropic Claude, and Meta AI are now actively researching products, comparing specs, and executing real-world checkouts on behalf of consumers.
The 4 Pillars of Modern AI Agent Readiness
To safely welcome buying agents while protecting server resources and user security, modern Joomla stores must implement four interoperable protocol standards:
| Standard / Protocol | Discovery Path | Purpose |
|---|---|---|
| llms.txt & llms-full.txt | /llms.txt |
Concise, markdown-formatted store summary and product directory for LLM context windows. |
| DNS-AID (DNS for AI Discovery) | /.well-known/dns-aid.json |
Decentralized DNS-based discovery allowing agents to locate store capabilities in 1 RTT lookup. |
| Personal Agent Protocol (PAP) | /.well-known/personal-agent.json |
Enterprise-grade handshake for consumer agents authorizing shopping actions and cart interactions. |
| Model Context Protocol (MCP) | /.well-known/mcp.json |
Standardized server tools allowing agents to query inventory, pricing, and stock in real time. |
1. Implementing llms.txt in Joomla
Published as an emerging open standard by Jeremy Howard (Answer.AI), llms.txt serves as an automated "robots.txt for AI understanding". It provides clean, token-efficient Markdown containing your store’s value propositions, active categories, shipping policies, and product URLs.
2. DNS for AI Discovery (DNS-AID)
Backed by the Linux Foundation and standardized via the IETF, DNS-AID enables AI agents to query your domain name system directly without crawling web pages. By publishing an _agent DNS TXT record pointing to your well-known manifest, agents instantly authenticate your store’s agent endpoint:
3. Personal Agent Protocol (PAP) Handshake
The Personal Agent Protocol (developed with Meta, Sierra, Shopify, and Stripe) governs how consumer personal assistants sign in and interact with your store. Rather than blindly executing form submissions, PAP establishes:
- Scoped Permissions: Read-only catalog access vs. authorized cart modifications.
- Human-in-the-Loop Safeguards: High-value orders require explicit human cryptographic biometric approval before capture.
- Content Negotiation: Responding with clean
application/jsonortext/markdownwhen requested by agents.
Frequently Asked Questions
What is AI Agent Readiness in e-commerce? ▼
AI Agent Readiness is the technical architecture that allows autonomous LLM agents (such as OpenAI Operator, ChatGPT Search, Claude, and Perplexity) to discover, inspect, and purchase products from an online store via machine-readable endpoints rather than scraping brittle HTML.
What is DNS for AI Discovery (DNS-AID)? ▼
DNS-AID is an open IETF specification backed by the Linux Foundation that utilizes standard DNS TXT and Service Binding (SVCB) records to allow AI agents to locate an online store's capability card and API endpoints in a single DNS query.
How does Personal Agent Protocol (PAP) protect merchants? ▼
The Personal Agent Protocol (PAP) provides a secure, cryptographically verifiable handshake between a consumer's personal AI agent and an enterprise store. It enforces user consent, prevents automated cart abuse, and ensures valid transaction liability.
Does FlexExt Store support llms.txt and MCP out of the box? ▼
Yes, FlexExt Store provides automated llms.txt feeds, full-catalog machine-readable JSON exports, Model Context Protocol (MCP) server endpoints, and HTTP Content Negotiation for Markdown responses.