Quick Answer: AAO (AI Agent Optimisation) is the practice of making your business discoverable and actionable to autonomous AI agents that browse, research, and transact on behalf of users. In Singapore, this involves publishing an /llms.txt file (a proposed community convention for LLMs), deploying connected JSON-LD schema graphs, and experimenting with emerging machine-readable interfaces such as WebMCP.
Why machine readability matters in Singapore
Singapore maintains one of the fastest rates of enterprise and public-sector AI adoption in the region. Singapore GovTech has deployed AI assistants (such as the Pair and Applier frameworks) to over 150,000 public sector officers (GovTech Singapore, 2024/2025). As autonomous agents in procurement, finance, and logistics begin conducting initial vendor screenings, domains that offer structured, machine-legible endpoints can be evaluated with greater fidelity than unstructured websites.
The three layers of agent readiness
Layer 1: llms.txt — Proposed Community Convention
/llms.txt is a proposed community convention for providing clean, markdown-formatted summaries of a website's core offerings, architecture, and contact information. While not a standardized W3C specification or a traditional search engine ranking factor, it provides a lightweight reference document for AI models and autonomous agents seeking structured brand facts without HTML overhead.
Layer 2: Connected JSON-LD Schema Graphs
Standardized Schema.org markup remains the universal semantic vocabulary recognized by Google, Bing, and generative AI crawlers. Deploying interconnected Organization, LocalBusiness, Service, and FAQPage schemas linked by persistent @id identifiers enables automated entity disambiguation across knowledge graphs.
Layer 3: WebMCP (Emerging Standard — Experimental)
The Model Context Protocol (MCP), open-sourced by Anthropic in late 2024, establishes a protocol for connecting AI models to external tools and data sources. WebMCP represents an experimental adaptation of this protocol to public web servers (via /.well-known/mcp.json and JSON endpoints). While experimental and not evaluated as a direct search ranking signal by search crawlers, it provides an architectural testing ground for direct agentic interactions.
Implementation Roadmap
- Audit your robots.txt to ensure AI user-agents (GPTBot, ClaudeBot, PerplexityBot) are permitted.
- Deploy comprehensive JSON-LD schema with verified entity references.
- Publish a compliant
/llms.txtat the domain root. - Run a diagnostic inspection via our Free AI Visibility Audit.