Methodology

Every product with a status other than dismissed: candidates awaiting review, cataloged and parked products, all included. A weekly engine discovers candidates from web search, Hacker News, Product Hunt, and GitHub, then a cheap model extracts facts from the homepage; a human curator pins, parks, or dismisses, and that decision is sticky, the scoring agent never relabels it. Deduplication matches on the normalized homepage URL first, then on a normalized name plus vendor domain. agent_fit and the five agent_* rubric scores are the scoring agent's advisory read, never a verdict; agent_steal lists up to three features worth stealing. dossier_summary, customers, and sources come from the machine's own merged research record, monotonically updated across homepage enrichment and deep dives; a deep dive is automatic for high-fit new entrants and available on demand for any product. review_note, raw_content, and fetch_error are the admin's working layer and never ship. Downloading requires an access key.

Schema

Column	Type	Definition
id	text	Stable UUID; the internal join key.
slug	text	Stable identifier; links to /tooling/<slug>.
name	text	Product name.
vendor	text	Vendor or maker, when known.
vendor_domain	text	Vendor's homepage domain, when known.
url	text	Product homepage, when known.
category	text	The curated category this product is filed under.
category_name	text	Display name of the category.
secondary_categories	text	Additional category slugs, joined with a semicolon.
one_liner	longtext	What the product does, in one sentence.
description	longtext	A fuller description of the product.
target_buyer	text	Who the product is sold to, joined with a semicolon.
deployment	text	Deployment options, joined with a semicolon.
pricing_model	enum	free, freemium, per_seat, usage, enterprise, or unknown, when known.
pricing_note	text	Free-text pricing detail.
maturity	enum	startup_early, startup_growth, scaleup, incumbent, big_tech, open_source_project, or unknown.
founded_year	number	Founding year, when reported.
hq	text	Headquarters, when reported.
funding_note	text	Latest reported funding, free text.
notable_customers	text	Named customers, joined with a semicolon.
integrations	text	Named integrations, joined with a semicolon.
compliance_claims	text	Compliance or certification claims, joined with a semicolon.
models_used	text	Underlying AI models named, joined with a semicolon.
features	text	Normalized feature tags, joined with a semicolon.
feed_url	text	Discovered RSS or Atom feed, when one exists.
changelog_url	text	Changelog page, when known.
github_repo	text	GitHub repository, when the product is open source.
status	enum	candidate, cataloged, parked, or dismissed. The human curator's decision; sticky.
pinned	number	1 when a human pinned this product so the agent never relabels it, 0 otherwise.
agent_fit	number	The scoring agent's 0 to 100 fit read for a regulated financial-services buyer. Advisory, not a verdict.
agent_relevance	number	Rubric dimension, 1 to 5.
agent_enterprise_readiness	number	Rubric dimension, 1 to 5.
agent_differentiation	number	Rubric dimension, 1 to 5.
agent_momentum	number	Rubric dimension, 1 to 5.
agent_build_difficulty	number	Rubric dimension, 1 to 5: how hard this would be to build in house.
agent_steal	text	Up to three features the agent flagged as worth stealing, joined with a semicolon.
agent_reason	longtext	The scoring agent's written rationale.
agent_model	text	Model that produced the score, when one exists.
dossier_summary	longtext	The machine's latest merged read of the product, when one exists.
customers	text	Customers recorded in the merged dossier, joined with a semicolon.
sources	text	Source URLs the dossier was built from, joined with a semicolon.
deep_dive_summary	longtext	Summary from the most recent deep dive, when one exists.
deep_dived_at	date	When the most recent deep dive ran (YYYY-MM-DD), when one exists.
origin	enum	How the product was found: tavily, hn, producthunt, github, enumeration, manual, or feed.
found_url	text	The URL discovery actually found, before resolving to a homepage.
first_seen	date	When the product first entered the catalog (YYYY-MM-DD).
last_seen	date	Most recent date the product was reconfirmed (YYYY-MM-DD).
enriched_by	text	Model that produced the homepage extraction, when one exists.
updated_at	date	When the row last changed (YYYY-MM-DD).

Tooling Monitor: products

Request access Unlock with an access key

/api/datasets/tooling-products · 49 columns · access key required

The AI Tooling Monitor catalog: every product the market scan has kept, curated categories plus emergent feature tags, the agent's fit read, and the machine's merged dossier and deep dive.

Methodology

Schema

Column	Type	Definition
id	text	Stable UUID; the internal join key.
slug	text	Stable identifier; links to /tooling/<slug>.
name	text	Product name.
vendor	text	Vendor or maker, when known.
vendor_domain	text	Vendor's homepage domain, when known.
url	text	Product homepage, when known.
category	text	The curated category this product is filed under.
category_name	text	Display name of the category.
secondary_categories	text	Additional category slugs, joined with a semicolon.
one_liner	longtext	What the product does, in one sentence.
description	longtext	A fuller description of the product.
target_buyer	text	Who the product is sold to, joined with a semicolon.
deployment	text	Deployment options, joined with a semicolon.
pricing_model	enum	free, freemium, per_seat, usage, enterprise, or unknown, when known.
pricing_note	text	Free-text pricing detail.
maturity	enum	startup_early, startup_growth, scaleup, incumbent, big_tech, open_source_project, or unknown.
founded_year	number	Founding year, when reported.
hq	text	Headquarters, when reported.
funding_note	text	Latest reported funding, free text.
notable_customers	text	Named customers, joined with a semicolon.
integrations	text	Named integrations, joined with a semicolon.
compliance_claims	text	Compliance or certification claims, joined with a semicolon.
models_used	text	Underlying AI models named, joined with a semicolon.
features	text	Normalized feature tags, joined with a semicolon.
feed_url	text	Discovered RSS or Atom feed, when one exists.
changelog_url	text	Changelog page, when known.
github_repo	text	GitHub repository, when the product is open source.
status	enum	candidate, cataloged, parked, or dismissed. The human curator's decision; sticky.
pinned	number	1 when a human pinned this product so the agent never relabels it, 0 otherwise.
agent_fit	number	The scoring agent's 0 to 100 fit read for a regulated financial-services buyer. Advisory, not a verdict.
agent_relevance	number	Rubric dimension, 1 to 5.
agent_enterprise_readiness	number	Rubric dimension, 1 to 5.
agent_differentiation	number	Rubric dimension, 1 to 5.
agent_momentum	number	Rubric dimension, 1 to 5.
agent_build_difficulty	number	Rubric dimension, 1 to 5: how hard this would be to build in house.
agent_steal	text	Up to three features the agent flagged as worth stealing, joined with a semicolon.
agent_reason	longtext	The scoring agent's written rationale.
agent_model	text	Model that produced the score, when one exists.
dossier_summary	longtext	The machine's latest merged read of the product, when one exists.
customers	text	Customers recorded in the merged dossier, joined with a semicolon.
sources	text	Source URLs the dossier was built from, joined with a semicolon.
deep_dive_summary	longtext	Summary from the most recent deep dive, when one exists.
deep_dived_at	date	When the most recent deep dive ran (YYYY-MM-DD), when one exists.
origin	enum	How the product was found: tavily, hn, producthunt, github, enumeration, manual, or feed.
found_url	text	The URL discovery actually found, before resolving to a homepage.
first_seen	date	When the product first entered the catalog (YYYY-MM-DD).
last_seen	date	Most recent date the product was reconfirmed (YYYY-MM-DD).
enriched_by	text	Model that produced the homepage extraction, when one exists.
updated_at	date	When the row last changed (YYYY-MM-DD).