AI Surface

This site is built for both humans and agents. The point is practical: the same public facts, offers, evidence, and representation rules are available as HTML, JSON, llms.txt, well-known metadata, and MCP.

Why this matters

AI systems are becoming part of how people search, compare, and operate. If those systems are going to talk about my work, I want them to use accurate, bounded, verifiable context instead of scraped guesses.

That is also how I approach client-facing AI work: useful tools, explicit permissions, clear evidence, and boring operational controls before hype.

Public machine-readable surfaces

Representation

/llms.txt and /prompt/represent_me.md tell agents how to summarize my work without resurrecting old positioning or overclaiming.

Structured data

/api/v1/profile.json, /api/v1/offers.json, and /api/v1/case_studies.json expose the same public facts used by the human pages.

Agent access

/.well-known/agent-card.json, /.well-known/ai.json, and /mcp/ make the profile queryable through bounded tools.

What this demonstrates

Designing AI-readable systems without making humans read a manifesto.
Separating public facts, private context, and retired surfaces.
Keeping agent tools bounded and aligned with published metadata.
Testing the public contract so stale content does not accidentally come back.

This is the same pattern I use for MCP/security integrations: clear scope, safe defaults, testable contracts, and evidence that survives handoff.

See services