A small-business guide to agent-ready infrastructure: what MCP servers, agent.json, llms.txt, and MCP manifests are; why agents are becoming a customer channel (Post Bridge, Amplitude, Lindy); how to scope and price an "agent-readiness" project.
A new kind of customer is arriving, and it doesn't have eyes. AI agents — inside ChatGPT, Claude, and coding tools — are starting to browse, compare, and buy on behalf of real people. In September 2026, solo founder Jack Friks reported a record signup day for his Post Bridge scheduling tool driven by ChatGPT referrals, and he ships an API, CLI, and MCP server so agents can post for users directly. Frank Lee's Amplitude team ships an MCP server that brings analytics into Claude, ChatGPT, Cursor, and Claude Code. Flo Crivello's Lindy added MCP support in a September mobile-first push. Caleb Peffer's Firecrawl launched Alexandria, a paid-data layer where contributors earn when agents use their data.
Summary drawn from the paper's executive summary (Human-N-Loop research series, October 2026 draft).
companion to the $97 AI-Readiness Audit; natural upsell into the $197 Fix-It and $297/mo tier.
Sources named in the research: Jack Friks, Frank Lee, Flo Crivello, Caleb Peffer, bot-paid-info-research.