# RafalAI

> RafalAI is a public educational reference for practical AI at work: workflows, source material, agents, MCP tools, machine-readable sites, search visibility, templates, evidence, review, and useful artifacts.

RafalAI is the author of the site's articles. RafalAI Labs runs the experiments behind them — starting with this site itself, which implements the WebMCP, machine-readable, and measurement patterns the articles describe. RafalAI explains how to make real work usable by AI without turning the site into a generic chatbot, prompt library, AI news feed, or one-tool demo. The site emphasizes practical workflow structure: start with the work, organize the source material, connect tools when useful, keep evidence visible, and review outputs before trusting, publishing, or sending them.

## Core Pages

- [Homepage](https://rafalai.com/): Public landing page for the RafalAI point of view, practical workflow stack, site/search/agent readiness themes, and Ask RafalAI intake form.
- [Articles](https://rafalai.com/articles/): Index of published RafalAI educational articles.
- [Practical AI Without The Theater](https://rafalai.com/articles/practical-ai-without-the-theater/): Article explaining the core thesis: useful AI starts with real work, source material, tools, templates, checks, useful artifacts, and human review.
- [Full LLM Context](https://rafalai.com/llms-full.txt): Expanded public context for summarizing the site.

## Public Site Files

- [Sitemap](https://rafalai.com/sitemap.xml): Canonical public URL list.
- [Robots](https://rafalai.com/robots.txt): Public crawl policy.
- [Article Feed](https://rafalai.com/feed.xml): RSS feed for published RafalAI articles.

## Public Interaction

- [Ask RafalAI](https://rafalai.com/#ask): Lightweight public intake form for practical AI, workflow, site, or educational questions. The form uses explicit labels, named fields, `toolparamdescription` attributes, `toolname="ask_rafalai"`, and a matching browser-native `document.modelContext.registerTool` registration when available.

## Crawl Policy and Measurement Notes

- Search crawling, model training, and user-directed retrieval are separate policy surfaces (for example Googlebot, `OAI-SearchBot` vs `GPTBot`, and `Claude-SearchBot` vs `ClaudeBot` vs `Claude-User`). RafalAI's robots.txt currently allows all of them. Browser agents acting for users are a fifth surface, governed by page semantics and declared WebMCP tools rather than robots.txt.
- ARIA labels, roles, and explicit states help browser agents interpret forms and controls the same way assistive technology does. RafalAI's public form uses visible labels, named fields, and described parameters.
- AI search visibility should be measured with first-party reporting: Google Search Console's generative-AI visibility views and referral tags such as `utm_source=chatgpt.com`, not single-run answer screenshots.
- Google states that AI features in Search require no special AI-only markup or files beyond normal Search eligibility.

## Topics

- Practical AI at work: research, drafting, analysis, reporting, planning, review, and repetitive knowledge work.
- Agents and MCP: using tools, files, APIs, forms, browsers, and approval points with clear boundaries.
- AI search and machine-readable sites: metadata, structured pages, `llms.txt`, WebMCP-style forms, SEO, GEO, source-aware writing, and crawlable public context.
- Workflow systems: prompts, skills, templates, checklists, design instructions, source material, and review loops.
- Human judgment: evidence, permissions, limits, and review before outputs are trusted or used.

## Optional

- [RafalAI Brand Assets](https://rafalai.com/brand/exports/og-image.png): Public Open Graph image and related visible brand assets used by the site.