From c041641634650c31e03c70dcad132fd94cb08e63 Mon Sep 17 00:00:00 2001 From: benj Date: Thu, 9 Apr 2026 16:10:08 +0800 Subject: Rework landing page with editorial layout, new demo, and LLM CTA MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Replaces the generic SaaS template look with a left-aligned editorial design — Fraunces serif headlines, monospace section markers (§ 01, § 02 …), hairline rules, and a quieter palette where the blue earns its place rather than splashing on every surface. The hero now leads with a real product demo (§ 01): one IRS 990 dataset shown in three consumption shapes — a single normalized JSON record, a JSONL bulk slice, and a token-efficient prose summary — instead of the old fake-macOS code window. Each shape carries source-link receipts back to the original filing, which doubles as a credibility signal and a working illustration of the "every record has a receipt" promise. § 03 introduces a coverage widget driven by data/catalog.toml: four auto-scrolling category windows (securities, government, regulatory, science) showing 60 hand-picked datasets with seamless CSS-only infinite-scroll loops, fade masks, hover-pause, and a prefers-reduced-motion fallback that converts the windows into manually scrollable lists. Adds a discoverable /llms.txt at the static root — a markdown-light, human-readable site description aimed at LLM agents — wired up via in , a comment in robots.txt, and a small mono "for agents · GET /llms.txt →" satellite link sitting directly under the hero CTA so any agent shallow-scraping the page hits the machine path immediately. Other changes: drops the XML format demo and the "promises" ledger section entirely, swaps hello@ for contact@ across the site, and renames the bottom CTA to a single-action "Ready when you are. → API Key" close. Co-Authored-By: Claude Opus 4.6 (1M context) --- landing/layouts/_default/baseof.html | 3 +- landing/layouts/index.html | 283 ++++++++++++++--------------------- landing/layouts/partials/footer.html | 13 +- landing/layouts/partials/header.html | 12 +- 4 files changed, 127 insertions(+), 184 deletions(-) (limited to 'landing/layouts') diff --git a/landing/layouts/_default/baseof.html b/landing/layouts/_default/baseof.html index 89a0bd7..bab1b56 100644 --- a/landing/layouts/_default/baseof.html +++ b/landing/layouts/_default/baseof.html @@ -17,10 +17,11 @@ - + + {{ partial "header.html" . }} diff --git a/landing/layouts/index.html b/landing/layouts/index.html index 231ff61..a08109e 100644 --- a/landing/layouts/index.html +++ b/landing/layouts/index.html @@ -1,213 +1,152 @@ {{ define "main" }}

- Clean data, on tap -

- Structured data,
- served simply. +

+ Structured data,
+ served simply.

Tidy Index delivers clean, well-organized datasets through a single API designed for humans, machines, and LLM agents alike.

- Request access - See how it works -

- -

- - - - GET /v1/datasets/cities -

{
-  "id": "cities",
-  "updated": "2026-04-09T08:14:00Z",
-  "records": [
-    { "name": "Lisbon", "population": 548703 },
-    { "name": "Kyoto",  "population": 1463723 },
-    { "name": "Quito",  "population": 2011388 }
-  ]
-}

+ + + Get an API Key + + See a record

- What is Tidy Index -

We’ve already done the cleanup.

- Scraping. Reformatting. Reconciling. Re-scraping when the source changes - shape. We do all of that — once, properly — so you don’t - have to do any of it. +

+ for agents + GET /llms.txt → + a machine-readable invitation. No form, no salesperson.

- Pick your shape -

One dataset. Every shape you need.

- Ask for JSON, get JSON. Ask for chunks, get chunks. Same data underneath — - no reformatting, no glue code. +

§ 01 · one dataset, three shapes

The same data, in whatever shape your job needs.

+ Below is the irs-990 dataset — the annual return + every U.S. nonprofit files with the IRS, normally distributed as a + sprawling PDF or wall of e-file XML. Tidy Index serves it three ways: + a single record for your API, a streaming bulk export for your data + team, or a token-efficient summary your agents can drop straight into + context.

{
+  "ein": "20-0049703",
+  "name": "Wikimedia Foundation, Inc.",
+  "fiscal_year": 2023,
+  "revenue_usd": 180249000,
+  "expenses_usd": 169106000,
+  "net_assets_usd": 254971000,
+  "source": "https://apps.irs.gov/app/eos/details/?ein=200049703"
+}

{"ein":"20-0049703","name":"Wikimedia Foundation, Inc.","revenue_usd":180249000}
+{"ein":"20-0097189","name":"Mozilla Foundation","revenue_usd":34019000}
+{"ein":"26-1544963","name":"Khan Academy, Inc.","revenue_usd":85436000}
+…

Wikimedia Foundation, Inc. (EIN 20-0049703) reported
+$180.25M in revenue against $169.11M in expenses on
+its 2023 Form 990, ending the year with $254.97M in
+net assets.
 
-      
-        
-          
-        
-        Streaming
-        Webhooks and server-sent events the moment something changes. No polling, no stale caches.
-      
+Source: https://apps.irs.gov/app/eos/details/?ein=200049703

- Who it’s for -

Same data. Three kinds of readers.

- People, programs, and agents all want different things from the same - dataset. Tidy Index gives each of them what they expect. +

§ 02 · what it is

Open data and licensed data, kept tidy.

+ Some of what we serve comes from the open web. The rest is licensed + from publishers, data vendors, and partners who’ve trusted us with + their private feeds. All of it lives behind the same API, with the same + stable schemas, the same stable IDs, and the same freshness guarantees + — and one key gets you every record in every shape your tools + (or your agents) prefer.

- -

For humans

- A real web UI. Skim a schema, preview a few rows, grab a snapshot, - and copy a working curl command without leaving the page. -

- -

For computers

- Boring REST, the way you like it. OpenAPI specs, semver, ETags, - idempotent reads, and rate limits we’ll actually tell you about. -

- -

For LLM agents

- Endpoints shaped like tool calls. Chunks shaped like context. Every - response carries its own provenance, so the model never has to guess - where the data came from. -

- Why Tidy Index -

Less wrangling. More building.

§ 03 · what we cover

The index, abridged.

+ A small slice of what’s already in the catalog. +

- -
- Curated, not crawled. -
A human looks at every dataset before it ships. We’d rather have ten that are right than ten thousand that are almost right.
-
-
- -
- Schemas that don’t move under you. -
Versioned endpoints, deprecation windows, and a changelog you can subscribe to. If something’s about to break, you’ll be the first to know.
-
-
- - - -
- Every record has a receipt. -
Each row links back to where it came from. Your auditors will love it. Your models will stop making things up.
+
+ {{ range $.Site.Data.catalog.categories }} +
+ {{ .name }} +
+
+ {{ range .items }}
{{ . }}
+ {{ end }}{{ range .items }} + {{ end }} +
-
- -
- One key, every format. -
JSON, XML, CSV, or LLM chunks. Flip a header — same data, different shape.
-
-

+ {{ end }} +

+ +

100+ datasets — and the next one we add is whichever one you ask for.

- -

Ready when you are.

- +

§ 04 · start

Ready when you are.

- diff --git a/landing/layouts/partials/header.html b/landing/layouts/partials/header.html index 5f6d85b..e7a9d40 100644 --- a/landing/layouts/partials/header.html +++ b/landing/layouts/partials/header.html @@ -8,13 +8,11 @@ - Tidy Index + Tidy Index - +

+ Private beta + API Key → +

-- cgit v1.2.3

+ Structured data, + served simply.

We’ve already done the cleanup.

One dataset. Every shape you need.

The same data, in whatever shape your job needs.

Streaming

Same data. Three kinds of readers.

Open data and licensed data, kept tidy.

For humans

For computers

For LLM agents

Less wrangling. More building.

The index, abridged.

Ready when you are.

Ready when you are.

+ Structured data,
+ served simply.