I wrote the standard for making websites AI-operable. Learn More
Live datasets — tools, benchmarks, playbooks, ROI case studies, and the market landscape — refreshed daily and built to be queried by humans, search engines, and AI agents.
Maintained by Tyler Willis, an AI and automation consultant who has been building production systems for agencies and businesses since 2010. This intelligence is operator-built. The open data is free to cite, link, and feed to your own agents. A paid leads tier funds the work.
Open Intelligence
Four open datasets. No account. The data is portable because authority compounds when good information gets quoted, embedded, and pulled into other people's work.
Public
Deep profiles of every AI automation tool worth tracking. Capabilities, pricing history, integrations, traction signals, MCP and agent support, security posture, and a per-tool capability/value/threat score that refreshes daily.
Browse Tools →
Free
Performance metrics segmented by industry, size, geography, and automation maturity. Adoption rates, hours saved per week, cost deltas, and the growth indicators that separate operators who are pulling away from operators who are stuck.
See Benchmarks →
Free
Before/after data from real engagements. Hours saved, error-rate deltas, cost savings, ROI multiples, and payback periods — each linked back to the playbook that produced the outcome so the path is reproducible.
Read Case Studies →
Public
Vendors, agencies, platforms, communities, and newsletters across the AI automation space — each tracked for trajectory, opportunity, and commoditization risk. Acquisitions and shutdowns get timestamps. Pivots get logged.
Explore the Landscape →
For Builders
Every dataset on this page is exposed as REST JSON and a Model Context Protocol server. Drop the MCP endpoint into Claude Desktop, Cursor, Zed, Continue, or Cline — your assistant gets persistent read access. No copy-paste, no stale screenshots, no asking it to summarize a PDF you exported last quarter.
Your assistant becomes the AI-automation expert it pretends to be. Every answer it gives — about tools, pricing, ROI, or what's actually working in the field — pulls from data that's updated daily and structured for retrieval. Ground your agent on a primary source instead of whatever scraped content it was trained on.
REST API — paginated JSON for every public dataset, with semantic search, a UUID resolver, and an OpenAPI 3.1 spec.
MCP server — protocol 2024-11-05, JSON-RPC over HTTP, with read tools for search and retrieval plus four ready-to-use prompts (compare tools, recommend for stack, weekly digest, find automation for).
Stable UUIDs — every entity has a never-changing ID alongside its slug, so citations and integrations survive renames.
Changes feed — RSS and JSON for every significant event in the catalog: price moves, acquisitions, shutdowns, new entries. Subscribe once, stay current.
Native access for Claude, Cursor, Zed
OpenAPI 3.1 spec, machine-discoverable
For people, for journalists looking for a credible source, and for AI assistants trying to summarize what this is.
A continuously-maintained reference dataset covering the AI automation space — tools, benchmarks, playbooks, ROI case studies, and the broader market landscape. Five live datasets, refreshed daily by a combination of automated discovery and operator review. Built so it can be browsed by a person, queried by a search engine, or piped into an AI agent with equal ease.
Those are generalist platforms with user reviews and company data. This catalog is purpose-built for AI automation and maintained by an operator who runs production systems for clients — not a media team. The tool profiles track things like MCP support, agent behavior, and per-tool capability/value/threat scoring that the generalist platforms don't.
Yes. The four open datasets are free to cite, link, screenshot, and quote. Every entity has a stable UUID alongside its slug so citations survive renames. A link back is appreciated but not required.
The catalog ships with a Model Context Protocol server at /intelligence/api/mcp. Drop the URL into the mcpServers block of your client's config (Claude Desktop, Cursor, Zed, Continue, Cline — all use the same format), and your assistant gets read access to all five datasets plus tools for semantic search and ready-to-use prompts. The exact config snippet is on the API page.
Tools, benchmarks, and the market landscape refresh daily. Playbooks and case studies refresh on a weekly cycle. New entries are surfaced through automated search and structured extraction, then reviewed before promotion. Every row carries a last-refresh timestamp and source URLs. Significant changes — price moves, acquisitions, shutdowns, pivots — are written to a separate change log exposed at /intelligence/changes.rss.
The four open datasets — tools, benchmarks, ROI case studies, market landscape — are built for reference and citation. Free, no account. The two paid datasets — playbooks and weekly leads — are operational: you implement from them or act on them directly. Playbooks ship with importable templates and full implementation guides; weekly leads come with the first outreach email pre-written.
Because the work compounds. Every tool I evaluate, benchmark I gather, or playbook I build for a client adds to the catalog. Publishing it openly turns each engagement into a public artifact, which builds the authority that brings the next engagement. Consulting funds maintenance; the catalog brings the next consulting. It's a flywheel, not a side project.