FAQ
Is this really free?
Yes. The agent is MIT-licensed. The course is published under Creative Commons Attribution. There's no email gate, no upsell, no paid tier. If it's useful, share it; if you find a bug, send a PR.
How does this compare to paid courses on directory websites?
Several paid directory-builder courses exist and are good — they give you case studies, video, and ongoing updates. This site teaches the methodology for free, in original wording, plus ships the open-source agent. If you want the creator-led perspective, buy a paid course alongside this. If you just want working code, take this.
Do I need a GPU?
No, but it helps. Without a GPU, LLM calls take 60–120 seconds each. With even a midrange GPU, ~10 seconds. The agent works either way; the difference is build speed. You can also point the agent at hosted APIs (OpenAI, Anthropic) instead of Ollama if you'd rather pay per call than buy hardware.
Why Astro instead of WordPress?
Static HTML output is faster, more secure, easier to deploy, and free to host on Cloudflare Pages. WordPress works for directories but needs managed hosting ($30+/month), backups, security patching, and database operations. Astro removes all of that. If you already love WordPress, the underlying methodology in the course applies regardless — only the build module changes.
Does this work outside the US?
Yes. OpenStreetMap covers the world, and Google indexes most countries. The agent's defaults assume US English; for other languages, change the LLM prompt language in config/default.yaml and use country-specific data sources where OSM is thin.
Will this still work in 2 years?
The methodology is durable; directories have been a viable web business since the late 1990s and the AI shift hasn't changed the fundamentals. The agent's specific integrations will need maintenance — Google's SERP HTML changes, Ahrefs's UI shifts, OSM tags evolve. That's why it's open source: future you (or anyone) can keep it current. The architecture is built to be patched, not rewritten.
What's the catch?
Time. Directories need 1–2 months of patient waiting after launch before Google ranks them. The agent compresses everything except the waiting. If you want overnight results, this isn't the business; if you want a calm, durable asset, it's a great fit.
Can I use this for client work?
Yes — MIT license. If you build directories for clients using this stack, do credit the project and the original experienced directory operators methodology somewhere visible. Otherwise, no obligations.
How do I report bugs or request features?
GitHub issues. The repo is github.com/mikee-ai/coldpitch. Pull requests welcome; the codebase is intentionally small and well-organized to make contribution easy.
Who runs this site?
Mike at mikee.ai. The site is hosted on Cloudflare Pages for free. There's no monetization on directory.mikee.ai itself — it's a public good.
Don't see your question? Open an issue on GitHub or DM @mikee_ai.