Website report
Visit sitehttps://thinkingmachines.ai/
Connectionism: Research Blog by Thinking Machines Lab
Audit Full discovery audit·Language English (en)·Response HTTP 200·Served by Cloudflare
Analysed Nov 9, 2026 · 15 categories · 96 passed · 23 warnings · 2 issues · 29 informational
Audit score
F52/100
Needs work
- Agentic
- 100
- SEO
- 52
What we found
2Needs a fix23Could be better96Looks good29Informational- 1.Needs a fix: Your page has no main heading. see checkThe main heading gives visitors and search engines a clear topic.
- 2.Could be better: Discovery copy is distinct across sampled pages: needs attention. see checkUseful, specific content gives visitors answers and a reason to stay.
- 3.Could be better: Your title only names the brand. see checkPeople need to know what you offer before they click.
- 4.Could be better: Share image dimensions declared: needs attention. see checkComplete link previews make your pages easier to recognize and share.
- 5.Could be better: Your headings skip levels. see checkA clear outline makes content easier to navigate, including with a screen reader.
Built with 2 technologies· detected stack
Read from the public source and response headers of the sampled pages.
How it fits together
AstroSite generator
CloudflareCDN / edge
- CloudflareCDN / edge
- AstroSite generator
A framework or CDN signal describes the public response, not the origin host or private backend.
Site checks
Every observation from the crawl, grouped by what it describes. Open a row for the evidence and the recommended fix.
Metadata3Could be better8Looks good
95/100Could be better: Title says what the site doesTitle is only the site name: "Thinking Machines Lab".
Could be better · Metadata · Review suggested
Observed evidence
Title is only the site name: "Thinking Machines Lab".
Why this matters
The title and description are the text someone reads in a search result before deciding whether to click. It is there, but weaker than this check expects.
Recommended fix
Add a short descriptor after the name, for example "Brand — what it does for whom". People who do not know the name yet search for what you do.
Could be better: Meta description lengthDescription is 53 characters, under this audit's 70-character review point (a heuristic, not a Google minimum).
Could be better · Metadata · Review suggested
Observed evidence
Description is 53 characters, under this audit's 70-character review point (a heuristic, not a Google minimum).
Why this matters
The title and description are the text someone reads in a search result before deciding whether to click. It is there, but weaker than this check expects.
Recommended fix
Add useful context if the summary omits the page purpose. Do not pad it or force keywords to reach a count.
Could be better: No meta keywords tagA meta keywords tag is present ("thinky thinkingmachines machine learning deep learning ai").
Could be better · Metadata · Review suggested
Observed evidence
A meta keywords tag is present ("thinky thinkingmachines machine learning deep learning ai").
Why this matters
The title and description are the text someone reads in a search result before deciding whether to click. It is there, but weaker than this check expects.
Recommended fix
You can remove it. Google has ignored meta keywords since 2009, so the tag does nothing for search and shows competitors the terms you target.
Looks good: Page title exists
Looks good · Metadata · Measured on your page
Observed evidence
Title found: "Thinking Machines Lab"
Why this matters
The title and description are the text someone reads in a search result before deciding whether to click. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Title length
Looks good · Metadata · Review suggested
Observed evidence
Title is 21 characters, inside this audit's 15–60 review range.
Why this matters
The title and description are the text someone reads in a search result before deciding whether to click. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Meta description exists
Looks good · Metadata · Measured on your page
Observed evidence
Meta description found: "Connectionism: Research Blog by Thinking Machines Lab"
Why this matters
The title and description are the text someone reads in a search result before deciding whether to click. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Show 5 more checks
Looks good: Titles unique across pages
Looks good · Metadata · Measured on your page
Observed evidence
All 10 indexable sampled pages have distinct titles.
Why this matters
The title and description are the text someone reads in a search result before deciding whether to click. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Homepage isn't noindexed
Looks good · Metadata · Measured on your page
Observed evidence
No blocking robots meta tag on the homepage.
Why this matters
The title and description are the text someone reads in a search result before deciding whether to click. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Character encoding declared
Looks good · Metadata · Measured on your page
Observed evidence
UTF-8 declared in the Content-Type header and a meta tag.
Why this matters
The title and description are the text someone reads in a search result before deciding whether to click. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: HTML5 doctype
Looks good · Metadata · Measured on your page
Observed evidence
HTML5 doctype declared.
Why this matters
The title and description are the text someone reads in a search result before deciding whether to click. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Valid metadata order in <head>
Looks good · Metadata · Measured on your page
Observed evidence
No invalid element interrupts metadata in <head>.
Why this matters
The title and description are the text someone reads in a search result before deciding whether to click. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Indexing16Looks good
100/100Looks good: Single canonical declaration
Looks good · Indexing · Measured on your page
Observed evidence
One canonical declaration found.
Why this matters
These rules decide whether the page can be indexed at all. One wrong directive hides good work. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Noindex and canonical agree
Looks good · Indexing · Measured on your page
Observed evidence
No conflicting noindex and cross-page canonical instructions.
Why this matters
These rules decide whether the page can be indexed at all. One wrong directive hides good work. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: No meta-refresh redirect
Looks good · Indexing · Measured on your page
Observed evidence
No meta-refresh redirect.
Why this matters
These rules decide whether the page can be indexed at all. One wrong directive hides good work. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: No soft-404 response
Looks good · Indexing · Measured on your page
Observed evidence
No soft-404 pattern on sampled HTTP 200 pages.
Why this matters
These rules decide whether the page can be indexed at all. One wrong directive hides good work. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Canonical target integrity
Looks good · Indexing · Measured on your page
Observed evidence
Canonical is a valid https URL. Its host is checked by the canonical consistency check.
Why this matters
These rules decide whether the page can be indexed at all. One wrong directive hides good work. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Sampled URLs have aligned indexability signals
Looks good · Indexing · Measured on your page
Observed evidence
Indexability signals align across 10 sampled URL(s).
Why this matters
These rules decide whether the page can be indexed at all. One wrong directive hides good work. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Show 10 more checks
Looks good: robots.txt exists
Looks good · Indexing · Measured on your page
Observed evidence
robots.txt responds with 200.
Why this matters
These rules decide whether the page can be indexed at all. One wrong directive hides good work. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: robots.txt doesn't block everything
Looks good · Indexing · Measured on your page
Observed evidence
The audited URL is allowed for Googlebot. No path rule matches this URL.
Why this matters
These rules decide whether the page can be indexed at all. One wrong directive hides good work. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Sitemap exists
Looks good · Indexing · Measured on your page
Observed evidence
Sitemap found at https://thinkingmachines.ai/sitemap.xml.
Why this matters
These rules decide whether the page can be indexed at all. One wrong directive hides good work. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Sitemap parses and lists URLs
Looks good · Indexing · Measured on your page
Observed evidence
Sitemap lists 42 same-origin URLs.
Why this matters
These rules decide whether the page can be indexed at all. One wrong directive hides good work. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Sitemap declared in robots.txt
Looks good · Indexing · Measured on your page
Observed evidence
robots.txt declares the sitemap.
Why this matters
These rules decide whether the page can be indexed at all. One wrong directive hides good work. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Canonical URL set
Looks good · Indexing · Measured on your page
Observed evidence
Canonical set: https://thinkingmachines.ai/
Why this matters
These rules decide whether the page can be indexed at all. One wrong directive hides good work. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Canonical is absolute and self-referencing
Looks good · Indexing · Measured on your page
Observed evidence
Canonical is absolute and points at this page.
Why this matters
These rules decide whether the page can be indexed at all. One wrong directive hides good work. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: www vs apex consistency
Looks good · Indexing · Measured on your page
Observed evidence
Hostname is used consistently (thinkingmachines.ai).
Why this matters
These rules decide whether the page can be indexed at all. One wrong directive hides good work. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: URLs are readable slugs
Looks good · Indexing · Measured on your page
Observed evidence
URL slugs are readable (29 sampled).
Why this matters
These rules decide whether the page can be indexed at all. One wrong directive hides good work. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Sampled pages expose search-access signals
Looks good · Indexing · Measured on your page
Observed evidence
No restrictive search-access signal on 10 sampled page(s). Actual indexing is not verified.
Why this matters
These rules decide whether the page can be indexed at all. One wrong directive hides good work. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Content1Could be better8Looks good2Informational
84/100Could be better: Discovery copy is distinct across sampled pagesRepeated discovery copy found: description "Connectionism: Research Blog by Thinking Machines Lab" on https://thinkingmachines.ai/ and https://thinkingmachines.ai/blog/
Could be better · Content · Measured on your page
Observed evidence
Repeated discovery copy found: description "Connectionism: Research Blog by Thinking Machines Lab" on https://thinkingmachines.ai/ and https://thinkingmachines.ai/blog/
Why this matters
This is the substance of the page. Placeholders and repeated copy make it read as unfinished. It is there, but weaker than this check expects.
Recommended fix
Give each content page a distinct description, H1, and opening copy. If the pages intentionally represent one resource, consolidate them with a canonical or redirect.
Affected pages (2)
- https://thinkingmachines.ai/
- https://thinkingmachines.ai/blog/
Not measured: ReadabilityFlesch reading ease ≈ 20 — very difficult.
Not measured · Content · Not established
Observed evidence
Flesch reading ease ≈ 20 — very difficult.
Why this matters
This observation could not be established from the available evidence. Resolve the collection limit or inspect the relevant source before deciding whether a change is needed.
Recommended fix
Shorten sentences and prefer plain words. Aim for 50–70 for a general audience.
Not applicable: Title and H1 agreeNeed both a <title> and an H1 to compare them.
Not applicable · Content · Measured on your page
Observed evidence
Need both a <title> and an H1 to compare them.
Why this matters
This is the substance of the page. Placeholders and repeated copy make it read as unfinished. This check does not apply to the features observed in this sample.
Recommended fix
Add whichever is missing — the metadata and headings checks name it — then this check confirms the two describe the same topic.
Looks good: Crawlable content in source HTML
Looks good · Content · Measured on your page
Observed evidence
No recognized empty JavaScript application shell on 10 sampled page(s). The entered page exposes 658 publisher word(s).
Why this matters
This is the substance of the page. Placeholders and repeated copy make it read as unfinished. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Homepage content volume
Looks good · Content · Measured on your page
Observed evidence
The audited page has about 658 words of main content.
Why this matters
This is the substance of the page. Placeholders and repeated copy make it read as unfinished. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Text-to-HTML ratio
Looks good · Content · Measured on your page
Observed evidence
Text-to-HTML ratio is 28.9%.
Why this matters
This is the substance of the page. Placeholders and repeated copy make it read as unfinished. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Show 5 more checks
Looks good: No placeholder text
Looks good · Content · Measured on your page
Observed evidence
No placeholder text detected on crawled pages.
Why this matters
This is the substance of the page. Placeholders and repeated copy make it read as unfinished. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: No starter-template leftovers
Looks good · Content · Measured on your page
Observed evidence
No starter-template artifacts detected.
Why this matters
This is the substance of the page. Placeholders and repeated copy make it read as unfinished. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Crawled pages aren't near-duplicates
Looks good · Content · Measured on your page
Observed evidence
No near-duplicate main content among 10 indexable sampled pages.
Why this matters
This is the substance of the page. Placeholders and repeated copy make it read as unfinished. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Outbound references
Looks good · Content · Measured on your page
Observed evidence
14 outbound link(s) present.
Why this matters
This is the substance of the page. Placeholders and repeated copy make it read as unfinished. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Structured dataNot measured
Not measuredCould be better: JSON-LD presentNo JSON-LD structured data on the homepage.
Could be better · Structured data · Review suggested
Observed evidence
No JSON-LD structured data on the homepage.
Why this matters
Structured data tells machines who published the page and what it describes. Broken markup is ignored outright. It is there, but weaker than this check expects.
Recommended fix
Add Organization and WebSite schema. They state who publishes the site and what it is called, which is what rich results and answer engines read first.
Could be better: Organization or WebSite schemaNo Organization, WebSite or Person schema found.
Could be better · Structured data · Review suggested
Observed evidence
No Organization, WebSite or Person schema found.
Why this matters
Structured data tells machines who published the page and what it describes. Broken markup is ignored outright. It is there, but weaker than this check expects.
Recommended fix
Add an identity node so search engines and AI assistants can attribute your content to a real entity.
Not applicable: JSON-LD parsesNo JSON-LD blocks were found, so there is no syntax to validate.
Not applicable · Structured data · Measured on your page
Observed evidence
No JSON-LD blocks were found, so there is no syntax to validate.
Why this matters
Structured data tells machines who published the page and what it describes. Broken markup is ignored outright. This check does not apply to the features observed in this sample.
Recommended fix
Add a <script type="application/ld+json"> block describing the page — Organization or WebSite is the usual starting point.
Not applicable: Content-specific schemaNo content-specific schema (Article, Product, FAQPage…) detected.
Not applicable · Structured data · Measured on your page
Observed evidence
No content-specific schema (Article, Product, FAQPage…) detected.
Why this matters
Structured data tells machines who published the page and what it describes. Broken markup is ignored outright. This check does not apply to the features observed in this sample.
Recommended fix
If you publish articles, products or FAQs, add the matching schema type — that's what makes rich results (stars, FAQs, prices) possible.
Headings1Needs a fix1Could be better3Looks good2Informational
12/100Needs a fix: H1 existsThe audited page has no <h1> (it starts at H2).
Needs a fix · Headings · Review suggestedCritical
Observed evidence
The audited page has no <h1> (it starts at H2).
Why this matters
Headings are the outline every parser and screen reader follows to find the answer. Right now the sampled pages don't provide it.
Recommended fix
Add one <h1> stating what the page is about, usually the hero headline. Search engines, screen readers and answer engines use it as the page's main heading.
Could be better: Heading levels don't skip1 heading level skip: H2 → H4 "Models".
Could be better · Headings · Review suggested
Observed evidence
1 heading level skip: H2 → H4 "Models".
Why this matters
Headings are the outline every parser and screen reader follows to find the answer. It is there, but weaker than this check expects.
Recommended fix
Use the next level down for each subsection (H2 under H1, H3 under H2). If a heading level was chosen for its font size, keep the level and change the styling instead.
Not applicable: Single H1No H1 exists to check for uniqueness.
Not applicable · Headings · Measured on your page
Observed evidence
No H1 exists to check for uniqueness.
Why this matters
Headings are the outline every parser and screen reader follows to find the answer. This check does not apply to the features observed in this sample.
Recommended fix
The H1 presence check covers the missing main heading.
Not applicable: H1 has real textNo H1 text exists to inspect.
Not applicable · Headings · Measured on your page
Observed evidence
No H1 text exists to inspect.
Why this matters
Headings are the outline every parser and screen reader follows to find the answer. This check does not apply to the features observed in this sample.
Recommended fix
The H1 presence check covers the missing main heading.
Looks good: Headings are descriptive
Looks good · Headings · Measured on your page
Observed evidence
11 headings, averaging 2.2 words — descriptive.
Why this matters
Headings are the outline every parser and screen reader follows to find the answer. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Content is sectioned with H2s
Looks good · Headings · Measured on your page
Observed evidence
5 H2 section(s) structure the page.
Why this matters
Headings are the outline every parser and screen reader follows to find the answer. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Element IDs are unique per document
Looks good · Headings · Measured on your page
Observed evidence
All inspected element IDs are unique within their documents.
Why this matters
Headings are the outline every parser and screen reader follows to find the answer. The inspected source provides this signal.
Recommended fix
Nothing to do here.
ImagesNot applicable
Not measuredNot applicable: Alt text coverageNo visible <img> elements on the audited page (hidden, presentational and 1×1 tracking images are not counted).
Not applicable · Images · Measured on your page
Observed evidence
No visible <img> elements on the audited page (hidden, presentational and 1×1 tracking images are not counted).
Why this matters
Alt text is what a screen reader announces, and what image search and AI systems read in place of the picture. This check does not apply to the features observed in this sample.
Recommended fix
Nothing to change here. Any image you add later needs alt text that describes it, so this check starts scoring then.
Not applicable: Width/height attributesNo visible <img> elements to measure on the audited page.
Not applicable · Images · Measured on your page
Observed evidence
No visible <img> elements to measure on the audited page.
Why this matters
Alt text is what a screen reader announces, and what image search and AI systems read in place of the picture. This check does not apply to the features observed in this sample.
Recommended fix
Nothing to change here. Give future images explicit width and height attributes so the browser can reserve their space.
Not applicable: Modern image formatsNo raster images with a recognizable file extension were found.
Not applicable · Images · Measured on your page
Observed evidence
No raster images with a recognizable file extension were found.
Why this matters
Alt text is what a screen reader announces, and what image search and AI systems read in place of the picture. This check does not apply to the features observed in this sample.
Recommended fix
Nothing to change here. If images arrive later, serve WebP or AVIF alongside PNG/JPEG.
Not applicable: Lazy loadingOnly 0 visible image(s) on the page, too few for lazy loading to change the initial load.
Not applicable · Images · Measured on your page
Observed evidence
Only 0 visible image(s) on the page, too few for lazy loading to change the initial load.
Why this matters
Alt text is what a screen reader announces, and what image search and AI systems read in place of the picture. This check does not apply to the features observed in this sample.
Recommended fix
Nothing to change here. Once the page carries several images, add loading="lazy" to the ones below the fold.
Links6Looks good
100/100Looks good: Internal links resolve
Looks good · Links · Measured on your page
Observed evidence
All 12 sampled internal links return < 400.
Why this matters
Links are how visitors and crawlers reach the rest of the site. A broken one is a dead end. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Anchor text is meaningful
Looks good · Links · Measured on your page
Observed evidence
38 of 38 links have descriptive text or labels.
Why this matters
Links are how visitors and crawlers reach the rest of the site. A broken one is a dead end. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: No dead placeholder links
Looks good · Links · Measured on your page
Observed evidence
No placeholder links on the audited page.
Why this matters
Links are how visitors and crawlers reach the rest of the site. A broken one is a dead end. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: External links use rel=noopener
Looks good · Links · Measured on your page
Observed evidence
All target="_blank" links carry rel="noopener".
Why this matters
Links are how visitors and crawlers reach the rest of the site. A broken one is a dead end. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Homepage links into the site
Looks good · Links · Measured on your page
Observed evidence
Homepage links to 11 distinct internal pages.
Why this matters
Links are how visitors and crawlers reach the rest of the site. A broken one is a dead end. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Same-page links have source targets
Looks good · Links · Measured on your page
Observed evidence
32 same-page fragment link(s) have source targets.
Why this matters
Links are how visitors and crawlers reach the rest of the site. A broken one is a dead end. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Mobile4Looks good2Informational
100/100Not applicable: Responsive image markupOnly 0 image(s) on the page, too few for responsive image markup to change what phones download.
Not applicable · Mobile · Measured on your page
Observed evidence
Only 0 image(s) on the page, too few for responsive image markup to change what phones download.
Why this matters
Google indexes the phone rendering, so that is the version being graded. This check does not apply to the features observed in this sample.
Recommended fix
Nothing to change here. When the page carries several images, give them srcset and sizes so phones fetch smaller files.
Not measured: No tiny inline font sizesNo sub-12px inline font sizes detected; stylesheet rules and rendered sizes were not measured.
Not measured · Mobile · Not established
Observed evidence
No sub-12px inline font sizes detected; stylesheet rules and rendered sizes were not measured.
Why this matters
This observation could not be established from the available evidence. Resolve the collection limit or inspect the relevant source before deciding whether a change is needed.
Recommended fix
Check computed font sizes in a browser across mobile widths.
Looks good: Viewport meta tag
Looks good · Mobile · Measured on your page
Observed evidence
Viewport configured: "width=device-width, initial-scale=1.0".
Why this matters
Google indexes the phone rendering, so that is the version being graded. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Pinch zoom not disabled
Looks good · Mobile · Measured on your page
Observed evidence
The viewport lets users zoom.
Why this matters
Google indexes the phone rendering, so that is the version being graded. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Interactive controls have source labels
Looks good · Mobile · Measured on your page
Observed evidence
473 source controls have detectable names. Browser accessibility and CSS visibility were not tested.
Why this matters
Google indexes the phone rendering, so that is the version being graded. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: ARIA references point to source elements
Looks good · Mobile · Measured on your page
Observed evidence
80 ARIA reference(s) resolve within their source documents. Rendered visibility was not tested.
Why this matters
Google indexes the phone rendering, so that is the version being graded. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Language2Looks good1Informational
100/100Not applicable: hreflang syntaxNo hreflang tags — fine for single-language sites.
Not applicable · Language · Measured on your page
Observed evidence
No hreflang tags — fine for single-language sites.
Why this matters
The declared language tells browsers and assistive tech how to read the page and who it is for. This check does not apply to the features observed in this sample.
Recommended fix
Only add hreflang if you publish the same content in multiple languages/regions.
Looks good: html lang attribute
Looks good · Language · Measured on your page
Observed evidence
Language declared: lang="en".
Why this matters
The declared language tells browsers and assistive tech how to read the page and who it is for. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Declared language matches content
Looks good · Language · Measured on your page
Observed evidence
Declared language (en) matches detected content language.
Why this matters
The declared language tells browsers and assistive tech how to read the page and who it is for. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Branding2Looks good2Informational
100/100Not measured: Web app manifestNo web manifest linked.
Not measured · Branding · Not established
Observed evidence
No web manifest linked.
Why this matters
This observation could not be established from the available evidence. Resolve the collection limit or inspect the relevant source before deciding whether a change is needed.
Recommended fix
Optional, but a manifest.json enables install prompts and better mobile integration.
Not measured: Theme colorNo theme-color meta tag.
Not measured · Branding · Not established
Observed evidence
No theme-color meta tag.
Why this matters
This observation could not be established from the available evidence. Resolve the collection limit or inspect the relevant source before deciding whether a change is needed.
Recommended fix
Add <meta name="theme-color"> to tint the mobile browser UI to your brand color. Small touch, feels polished.
Looks good: Favicon
Looks good · Branding · Measured on your page
Observed evidence
Favicon declared via <link rel="icon"> (/images/favicon-32x32.png).
Why this matters
These icons are what appear in browser tabs, bookmarks and search results. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Apple touch icon
Looks good · Branding · Measured on your page
Observed evidence
apple-touch-icon is declared.
Why this matters
These icons are what appear in browser tabs, bookmarks and search results. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Performance
Measured from the audited response and its markup. These are source observations, not field Core Web Vitals.
| Measure | This site | Reading |
|---|---|---|
| Web fonts don't block text | — | Informational: not applicableNo Google Fonts stylesheet links detected. |
| Server response time | 120 ms | Looks good: goodThe audit received response headers after 120 ms from one test location. |
| HTML payload size | 66 KB | Looks good: goodThe page's HTML is a lean 66 KB uncompressed. |
| No render-blocking head scripts | — | Looks good: goodNo synchronous render-blocking scripts in <head>. |
| Inline styles are bounded | 48.5 KB | Looks good: goodInline styles total 48.5 KB. |
| Script/stylesheet count | — | Looks good: good4 total script/style resources — reasonable. |
| HTML served compressed | — | Looks good: goodHTML is compressed with gzip. |
| Iframe usage | — | Looks good: goodNo iframes. |
| Source HTML weight | 65.7 KB | Informational: measured source, not transferred bytes65.7 KB of HTML on the audited page |
From a single automated fetch of the audited page. A quick reading, not a substitute for Lighthouse or field data.
Security
HTTPS, transport security and the response headers the audited pages send.
Could be better: HSTS headerNo active Strict-Transport-Security max-age was observed.
Could be better · Security · Review suggested
Observed evidence
No active Strict-Transport-Security max-age was observed.
Why this matters
HTTPS protects what visitors send you, and response headers limit what a hostile page can do in their browser. It is there, but weaker than this check expects.
Recommended fix
Add HSTS (e.g. max-age=63072000; includeSubDomains) so browsers never even try HTTP after the first visit.
Could be better: Frame embedding protectionNo frame-embedding restriction was observed (neither CSP frame-ancestors nor X-Frame-Options).
Could be better · Security · Review suggested
Observed evidence
No frame-embedding restriction was observed (neither CSP frame-ancestors nor X-Frame-Options).
Why this matters
HTTPS protects what visitors send you, and response headers limit what a hostile page can do in their browser. It is there, but weaker than this check expects.
Recommended fix
Send Content-Security-Policy: frame-ancestors 'self' (or 'none', or a list of partner origins), or X-Frame-Options: SAMEORIGIN. It stops other sites from framing your pages to trick visitors into clicking (clickjacking).
Could be better: Content Security PolicyNo Content-Security-Policy detected.
Could be better · Security · Review suggested
Observed evidence
No Content-Security-Policy detected.
Why this matters
HTTPS protects what visitors send you, and response headers limit what a hostile page can do in their browser. It is there, but weaker than this check expects.
Recommended fix
Configure an enforced policy for script and resource origins; use report-only monitoring to prepare changes safely.
Could be better: Browser feature permissionsNo explicit restriction on sensitive browser features was observed.
Could be better · Security · Review suggested
Observed evidence
No explicit restriction on sensitive browser features was observed.
Why this matters
HTTPS protects what visitors send you, and response headers limit what a hostile page can do in their browser. It is there, but weaker than this check expects.
Recommended fix
Disable unused camera, microphone, or geolocation access using Permissions-Policy.
Not applicable: Password forms avoid URL submissionNo enabled, named password inputs in sampled HTML.
Not applicable · Security · Measured on your page
Observed evidence
No enabled, named password inputs in sampled HTML.
Why this matters
HTTPS protects what visitors send you, and response headers limit what a hostile page can do in their browser. This check does not apply to the features observed in this sample.
Recommended fix
Review authentication forms separately when applicable.
Looks good: Site serves over HTTPS
Looks good · Security · Measured on your page
Observed evidence
Site serves over HTTPS.
Why this matters
HTTPS protects what visitors send you, and response headers limit what a hostile page can do in their browser. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: HTTP redirects to HTTPS
Looks good · Security · Measured on your page
Observed evidence
http:// permanently redirects to https://./
Why this matters
HTTPS protects what visitors send you, and response headers limit what a hostile page can do in their browser. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Referrer policy
Looks good · Security · Measured on your page
Observed evidence
Referrer-Policy is strict-origin-when-cross-origin.
Why this matters
HTTPS protects what visitors send you, and response headers limit what a hostile page can do in their browser. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: X-Content-Type-Options
Looks good · Security · Measured on your page
Observed evidence
nosniff is set.
Why this matters
HTTPS protects what visitors send you, and response headers limit what a hostile page can do in their browser. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: No mixed content
Looks good · Security · Measured on your page
Observed evidence
No insecure subresources detected.
Why this matters
HTTPS protects what visitors send you, and response headers limit what a hostile page can do in their browser. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Server technology disclosure
Looks good · Security · Measured on your page
Observed evidence
No X-Powered-By disclosure.
Why this matters
HTTPS protects what visitors send you, and response headers limit what a hostile page can do in their browser. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Declared form submission targets
Looks good · Security · Measured on your page
Observed evidence
1 declared form(s) submit over HTTPS to this site. Runtime submission behavior is untested.
Why this matters
HTTPS protects what visitors send you, and response headers limit what a hostile page can do in their browser. The inspected source provides this signal.
Recommended fix
Nothing to do here.
AI answer readiness
Retrieval signals an answer engine relies on: crawler access, clear entities and answers it can quote. These checks do not measure citations or traffic.
Could be better: llms.txtNo /llms.txt file.
Could be better · AI answer readiness · Review suggested
Observed evidence
No /llms.txt file.
Why this matters
This is what an AI assistant relies on when deciding whether it can quote your page. It is there, but weaker than this check expects.
Recommended fix
Optional: publish an llms.txt index of useful pages. This proposal is not a requirement for Google AI features, and its absence does not reduce the core score.
Could be better: FAQ/QA schemaNo FAQPage or QAPage schema on any crawled page.
Could be better · AI answer readiness · Review suggested
Observed evidence
No FAQPage or QAPage schema on any crawled page.
Why this matters
This is what an AI assistant relies on when deciding whether it can quote your page. It is there, but weaker than this check expects.
Recommended fix
If the page answers recurring questions, use visible question-and-answer content and matching FAQPage markup. QAPage is for a single question with user-submitted answers. Google FAQ rich results are restricted to eligible authoritative government and health sites; markup does not guarantee AI citations.
Could be better: Question-style headingsOnly 3% of 223 H2/H3 headings are question-shaped.
Could be better · AI answer readiness · Review suggested
Observed evidence
Only 3% of 223 H2/H3 headings are question-shaped.
Why this matters
This is what an AI assistant relies on when deciding whether it can quote your page. It is there, but weaker than this check expects.
Recommended fix
Phrase key subheadings as the questions users actually ask ("How much does X cost?"). This improves the measured question-heading signal; the audit does not guarantee citations.
Could be better: Freshness signalsNo dateModified schema or visible 'updated' dates found.
Could be better · AI answer readiness · Review suggested
Observed evidence
No dateModified schema or visible 'updated' dates found.
Why this matters
This is what an AI assistant relies on when deciding whether it can quote your page. It is there, but weaker than this check expects.
Recommended fix
For content where dates matter, show the actual update date and mirror it in dateModified. Do not invent dates or change them without a substantive update. Missing dates alone do not prove stale content.
Could be better: Question headings have concise answers2/6 question-shaped heading(s) have a 20–500 character paragraph immediately after them.
Could be better · AI answer readiness · Measured on your page
Observed evidence
2/6 question-shaped heading(s) have a 20–500 character paragraph immediately after them.
Why this matters
This is what an AI assistant relies on when deciding whether it can quote your page. It is there, but weaker than this check expects.
Recommended fix
Put a direct 20–500 character answer paragraph immediately after each question heading so retrieval systems can quote it cleanly.
Affected pages (3)
- https://thinkingmachines.ai/legal/privacy/
- https://thinkingmachines.ai/tinker/
- https://thinkingmachines.ai/security/disclosure-policy/
Could be better: Machine-readable provenance signalsOnly 2 of 4 provenance signal types were found: date, author.
Could be better · AI answer readiness · Measured on your page
Observed evidence
Only 2 of 4 provenance signal types were found: date, author.
Why this matters
This is what an AI assistant relies on when deciding whether it can quote your page. It is there, but weaker than this check expects.
Recommended fix
Add the missing author, date, breadcrumb, or contact signal so machines can identify provenance and site structure.
Affected pages (4)
- https://thinkingmachines.ai/legal/privacy/
- https://thinkingmachines.ai/tinker/
- https://thinkingmachines.ai/legal/terms/
- https://thinkingmachines.ai/training-data-documentation/
Could be better: Markdown copies of pagesNo markdown copy answered at /index.md or /legal/privacy/index.md or /index.html.md.
Could be better · AI answer readiness · Measured on your page
Observed evidence
No markdown copy answered at /index.md or /legal/privacy/index.md or /index.html.md.
Why this matters
This is what an AI assistant relies on when deciding whether it can quote your page. It is there, but weaker than this check expects.
Recommended fix
Publish a clean markdown copy of each key page at /page.md (or /page.html.md). It strips navigation and scripts, so a retriever spends its budget on your content.
Affected pages (2)
- https://thinkingmachines.ai/
- https://thinkingmachines.ai/legal/privacy/
Not measured: llms-full.txtNo /llms-full.txt.
Not measured · AI answer readiness · Not established
Observed evidence
No /llms-full.txt.
Why this matters
This observation could not be established from the available evidence. Resolve the collection limit or inspect the relevant source before deciding whether a change is needed.
Recommended fix
Optional companion to llms.txt containing your full content as one markdown file — useful for sites whose pages are JS-heavy.
Not measured: Citable facts density6 concrete figures across 711 words.
Not measured · AI answer readiness · Not established
Observed evidence
6 concrete figures across 711 words.
Why this matters
This observation could not be established from the available evidence. Resolve the collection limit or inspect the relevant source before deciding whether a change is needed.
Recommended fix
Good — keep facts specific and current, and date-stamp them.
Not measured: Description reads as a liftable summaryDescription is short or non-declarative: "Connectionism: Research Blog by Thinking Machines Lab…"
Not measured · AI answer readiness · Not established
Observed evidence
Description is short or non-declarative: "Connectionism: Research Blog by Thinking Machines Lab…"
Why this matters
This observation could not be established from the available evidence. Resolve the collection limit or inspect the relevant source before deciding whether a change is needed.
Recommended fix
Prefer a complete declarative sentence over slogans or questions — it should stand alone when quoted out of context.
Not measured: llms.txt links resolve to this siteNo readable llms.txt was available to inspect for Markdown links.
Not measured · AI answer readiness · Not established
Observed evidence
No readable llms.txt was available to inspect for Markdown links.
Why this matters
This observation could not be established from the available evidence. Resolve the collection limit or inspect the relevant source before deciding whether a change is needed.
Recommended fix
Publish a readable llms.txt, then link its important pages with same-origin Markdown URLs.
Not measured: llms.txt follows the specNo readable /llms.txt, so its structure cannot be checked.
Not measured · AI answer readiness · Not established
Observed evidence
No readable /llms.txt, so its structure cannot be checked.
Why this matters
This observation could not be established from the available evidence. Resolve the collection limit or inspect the relevant source before deciding whether a change is needed.
Recommended fix
Publish llms.txt first — the llms.txt check covers that — then this one verifies the structure.
Not applicable: Accept: text/markdown negotiationThe homepage answered Accept: text/markdown with text/html; charset=utf-8.
Not applicable · AI answer readiness · Measured on your page
Observed evidence
The homepage answered Accept: text/markdown with text/html; charset=utf-8.
Why this matters
This is what an AI assistant relies on when deciding whether it can quote your page. This check does not apply to the features observed in this sample.
Recommended fix
Optional: return a stripped markdown body when the request asks for text/markdown. Some hosts and CDNs do this for you; a plain HTML answer is not a defect.
Not applicable: Advertised machine interfaces are inspectableNo OpenAPI interface was advertised by sampled HTML.
Not applicable · AI answer readiness · Measured on your page
Observed evidence
No OpenAPI interface was advertised by sampled HTML.
Why this matters
This is what an AI assistant relies on when deciding whether it can quote your page. This check does not apply to the features observed in this sample.
Recommended fix
An API is optional. Add a service-desc link if you already publish a machine-readable API specification.
Looks good: AI crawler access
Looks good · AI answer readiness · Measured on your page
Observed evidence
The audited URL is allowed for evaluated answer-search crawlers.
Why this matters
This is what an AI assistant relies on when deciding whether it can quote your page. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Semantic HTML structure
Looks good · AI answer readiness · Measured on your page
Observed evidence
Semantic structure present: main×1, article×1, nav×2, header×2, footer×1.
Why this matters
This is what an AI assistant relies on when deciding whether it can quote your page. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Answers lead the section
Looks good · AI answer readiness · Measured on your page
Observed evidence
Paragraphs get to the point without preamble.
Why this matters
This is what an AI assistant relies on when deciding whether it can quote your page. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Content is chunkable
Looks good · AI answer readiness · Measured on your page
Observed evidence
Paragraphs average 33 words, lists present — chunk-friendly.
Why this matters
This is what an AI assistant relies on when deciding whether it can quote your page. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Definitional opening
Looks good · AI answer readiness · Measured on your page
Observed evidence
Opening paragraph reads as a definition.
Why this matters
This is what an AI assistant relies on when deciding whether it can quote your page. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Cites external sources
Looks good · AI answer readiness · Measured on your page
Observed evidence
145 outbound reference(s) across crawled pages.
Why this matters
This is what an AI assistant relies on when deciding whether it can quote your page. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Consistent brand entity
Looks good · AI answer readiness · Measured on your page
Observed evidence
Brand declared: og:site_name "Thinking Machines Lab".
Why this matters
This is what an AI assistant relies on when deciding whether it can quote your page. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Entity identity agrees across signals
Looks good · AI answer readiness · Measured on your page
Observed evidence
The site is consistently named "Thinking Machines Lab" and 2 identity URL(s) stay on thinkingmachines.ai.
Why this matters
This is what an AI assistant relies on when deciding whether it can quote your page. The inspected source provides this signal.
Recommended fix
Nothing to do here.
AdSense readinessOptional
Publisher-policy signals, for sites that plan to carry ads. These checks do not affect the overall score or grade, and they do not measure or guarantee AdSense approval.
On its own, this site scores 32/100 for publisher readiness. Ignore this section if you do not run ads.
Needs a fix: Latest article names an author and a dateThe newest article shows neither an author nor a publication date.
Needs a fix · AdSense policy · Measured on your pageCritical
Observed evidence
The newest article shows neither an author nor a publication date.
Why this matters
A publisher review looks for this before it accepts a site. Right now the sampled pages don't provide it.
Recommended fix
Put a named author and a visible date on every article, and repeat both in Article JSON-LD. Reviewers look for a person who stands behind the page.
Affected pages (1)
- https://thinkingmachines.ai/blog/defeating-nondeterminism-in-llm-inference/
Could be better: Privacy policy advertising disclosuresFetched privacy policy lacks advertising or Google disclosure signals.
Could be better · AdSense policy · Published platform ruleCritical
Observed evidence
Fetched privacy policy lacks advertising or Google disclosure signals.
Why this matters
A publisher review looks for this before it accepts a site. It is there, but weaker than this check expects.
Recommended fix
Describe data collection, sharing, and cookie use by Google and other third-party ad vendors.
Affected pages (1)
- https://thinkingmachines.ai/legal/privacy/
Could be better: Sampled publisher-content coverage3/8 editorial pages expose fewer than 250 publisher words; those 8 pages hold 8734 publisher words in total.
Could be better · AdSense policy · Review suggestedCritical
Observed evidence
3/8 editorial pages expose fewer than 250 publisher words; those 8 pages hold 8734 publisher words in total. 2 of the 10 sampled pages are support or tool pages and were excluded from this check.
Why this matters
A publisher review looks for this before it accepts a site. It is there, but weaker than this check expects.
Recommended fix
Review whether each page fully serves its purpose. Expand incomplete editorial content with useful information; keep concise navigation and tool pages when they help visitors.
Affected pages (3)
- https://thinkingmachines.ai/inkling/
- https://thinkingmachines.ai/blog/
- https://thinkingmachines.ai/news/
Could be better: Contact methodNo contact page or mailto link found.
Could be better · AdSense policy · Review suggested
Observed evidence
No contact page or mailto link found.
Why this matters
A publisher review looks for this before it accepts a site. It is there, but weaker than this check expects.
Recommended fix
Add a working contact route or visible email so users can reach the publisher. This is a trust heuristic, not a published approval rule.
Not measured: First-party contact emailNo public contact email was found in the sampled home, About, or Contact pages.
Not measured · AdSense policy · Review suggested
Observed evidence
No public contact email was found in the sampled home, About, or Contact pages.
Why this matters
This observation could not be established from the available evidence. Resolve the collection limit or inspect the relevant source before deciding whether a change is needed.
Recommended fix
A public email is optional. The separate contact-method check covers whether readers have a way to reach the publisher.
Not measured: ads.txtNo ads.txt yet.
Not measured · AdSense policy · Not established
Observed evidence
No ads.txt yet.
Why this matters
This observation could not be established from the available evidence. Resolve the collection limit or inspect the relevant source before deciding whether a change is needed.
Recommended fix
When AdSense provides your publisher entry, publish it exactly at /ads.txt and monitor authorization status in your account.
Not measured: Domain age & historyDomain age isn't visible from the site itself.
Not measured · AdSense policy · Not established
Observed evidence
Domain age isn't visible from the site itself.
Why this matters
This observation could not be established from the available evidence. Resolve the collection limit or inspect the relevant source before deciding whether a change is needed.
Recommended fix
Do not use age as a content-quality proxy. Verify ownership and account eligibility in AdSense; Google's general eligibility page publishes no domain-age threshold.
Looks good: Privacy policy page
Looks good · AdSense policy · Measured on your page
Observed evidence
Privacy policy found: https://thinkingmachines.ai/legal/privacy/
Why this matters
A publisher review looks for this before it accepts a site. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Terms page
Looks good · AdSense policy · Measured on your page
Observed evidence
Terms page found: https://thinkingmachines.ai/legal/terms/
Why this matters
A publisher review looks for this before it accepts a site. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: About page
Looks good · AdSense policy · Measured on your page
Observed evidence
About page found: /blog/the-future-worth-building-is-human/
Why this matters
A publisher review looks for this before it accepts a site. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Thin page share
Looks good · AdSense policy · Review suggested
Observed evidence
8/8 editorial pages clear the audit's publisher-content triage threshold. 2 of the 10 sampled pages are support or tool pages and were excluded from this check.
Why this matters
A publisher review looks for this before it accepts a site. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: No ads on non-content screens
Looks good · AdSense policy · Published platform rule
Observed evidence
No Google ad code detected on sampled non-content screens.
Why this matters
A publisher review looks for this before it accepts a site. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Tool pages expose explanatory content
Looks good · AdSense policy · Review suggested
Observed evidence
All 10 sampled tool-style screens include crawlable explanation.
Why this matters
A publisher review looks for this before it accepts a site. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Publisher content differs across URLs
Looks good · AdSense policy · Review suggested
Observed evidence
No high publisher-content repetition among 9 substantial sampled page(s).
Why this matters
A publisher review looks for this before it accepts a site. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Publisher content exists in source HTML
Looks good · AdSense policy · Measured on your page
Observed evidence
No empty client-rendered publisher shell in the crawl sample.
Why this matters
A publisher review looks for this before it accepts a site. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Article-style content exists
Looks good · AdSense policy · Measured on your page
Observed evidence
Article-style content detected.
Why this matters
A publisher review looks for this before it accepts a site. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: No "coming soon" or placeholder pages
Looks good · AdSense policy · Measured on your page
Observed evidence
No "coming soon", "under construction" or lorem-ipsum text on crawled pages.
Why this matters
A publisher review looks for this before it accepts a site. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Site language is AdSense-supported
Looks good · AdSense policy · Measured on your page
Observed evidence
Language "en" is AdSense-supported.
Why this matters
A publisher review looks for this before it accepts a site. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Policy pages linked site-wide
Looks good · AdSense policy · Measured on your page
Observed evidence
All 10 crawled pages link to the privacy policy.
Why this matters
A publisher review looks for this before it accepts a site. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Show 4 more checks
Looks good: Existing ad scripts
Looks good · AdSense policy · Measured on your page
Observed evidence
No competing ad scripts on the page.
Why this matters
A publisher review looks for this before it accepts a site. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Latest article has real depth
Looks good · AdSense policy · Measured on your page
Observed evidence
The newest article carries 3284 words of body text.
Why this matters
A publisher review looks for this before it accepts a site. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Looks good: Latest article reads as edited work
Looks good · AdSense policy · Review suggested
Observed evidence
Style pattern score 4/100 on the newest article: 1 paragraph(s) opening with a filler transition. These are style patterns in the sampled text, not a determination of how the page was written.
Why this matters
A publisher review looks for this before it accepts a site. The inspected source provides this signal.
Recommended fix
Nothing to do here.
Hosting and infrastructure
Public endpoint facts observed at audit time. A CDN or proxy can conceal the origin infrastructure.
| Hostname | thinkingmachines.ai |
|---|---|
| CDN / edge | Cloudflare |
| IP addresses | 104.18.1.136, 104.18.0.136, 2606:4700::6812:88, 2606:4700::6812:188 |
| DNS A | 104.18.0.136 · 104.18.1.136 |
| DNS AAAA | 2606:4700::6812:88 · 2606:4700::6812:188 |
| DNS NS | alexa.ns.cloudflare.com · kellen.ns.cloudflare.com |
| DNS MX | 1 aspmx.l.google.com · 10 alt3.aspmx.l.google.com · 10 alt4.aspmx.l.google.com · 5 alt1.aspmx.l.google.com · 5 alt2.aspmx.l.google.com |
| DNS CAA | {"critical":0,"type":"CAA","issuewild":"digicert.com; cansignhttpexchanges=yes"} · {"critical":0,"type":"CAA","issue":"digicert.com; cansignhttpexchanges=yes"} · {"critical":0,"type":"CAA","issuewild":"sectigo.com"} · {"critical":0,"type":"CAA","issue":"comodoca.com"} · {"critical":0,"type":"CAA","issuewild":"pki.goog; cansignhttpexchanges=yes"} · {"critical":0,"type":"CAA","issue":"ssl.com"} · {"critical":0,"type":"CAA","issue":"sectigo.com"} · {"critical":0,"type":"CAA","issuewild":"ssl.com"} · {"critical":0,"type":"CAA","issue":"pki.goog; cansignhttpexchanges=yes"} · {"critical":0,"type":"CAA","issue":"letsencrypt.org"} · {"critical":0,"type":"CAA","issuewild":"comodoca.com"} · {"critical":0,"type":"CAA","iodef":"mailto:security@thinkingmachines.ai"} · {"critical":0,"type":"CAA","issuewild":"letsencrypt.org"} |
| TLS | TLSv1.3TLS_AES_256_GCM_SHA384 |
| Certificate issuer | Google Trust Services / WE1 |
| Certificate valid until | Jan 27 18:08:36 2027 GMT |
| Certificate validation | Passed |
Website anatomy and diagnostics
Explore links, resources, crawler access and site integrations.
Page architecture
Choose a page to explore its details and connections.
Page content
658 publisher words
Page structure
html content
Observed discovery
Depth 0 · 9 incoming links
HTTP 200 · 65.7 KB source
Source interactions
32 controls · 0 unnamed
3 script / 1 stylesheet references
- Canonical declaration
- https://thinkingmachines.ai/
- Schema types
- None observed
- Outgoing links in sampled graph
- /inkling/ · /tinker/ · /blog/ · /news/ · /news/inkling-2-small/ · /news/introducing-inkling/ · /blog/interaction-models/ · /blog/the-future-worth-building-is-human/ · /legal/terms/ · /legal/privacy/ · /security/disclosure-policy/
Routes: / · /legal · /inkling · /tinker · /blog · /news · /security · /model-card · /training-data-documentation
Repeated metadata in the sample
Same description: connectionism: research blog by thinking machines lab
/ · /blog/
Resource origins 2
Resources associated with your website. External resources may belong to the same organization.
- https://thinkingmachines.ai/stylesheet, script, image · 160 references · same origin
- https://static.cloudflareinsights.com/script · 10 references · cross origin
Crawler policy by page
Review crawler access for your pages. Search, training and user-triggered services have different purposes; declared access does not guarantee discovery.
| Crawler / purpose | Page | Policy | Matched directive |
|---|---|---|---|
| Googlebotsearch | / | allowed | Declared policy for this page. |
| Googlebotsearch | /legal/privacy/ | allowed | Declared policy for this page. |
| Googlebotsearch | /inkling/ | allowed | Declared policy for this page. |
| Googlebotsearch | /tinker/ | allowed | Declared policy for this page. |
| Googlebotsearch | /blog/ | allowed | Declared policy for this page. |
| Googlebotsearch | /news/ | allowed | Declared policy for this page. |
| Googlebotsearch | /legal/terms/ | allowed | Declared policy for this page. |
| Googlebotsearch | /security/disclosure-policy/ | allowed | Declared policy for this page. |
| Googlebotsearch | /model-card/inkling-2-small/ | allowed | Declared policy for this page. |
| Googlebotsearch | /training-data-documentation/ | allowed | Declared policy for this page. |
| bingbotsearch | / | allowed | Declared policy for this page. |
| bingbotsearch | /legal/privacy/ | allowed | Declared policy for this page. |
| bingbotsearch | /inkling/ | allowed | Declared policy for this page. |
| bingbotsearch | /tinker/ | allowed | Declared policy for this page. |
| bingbotsearch | /blog/ | allowed | Declared policy for this page. |
| bingbotsearch | /news/ | allowed | Declared policy for this page. |
| bingbotsearch | /legal/terms/ | allowed | Declared policy for this page. |
| bingbotsearch | /security/disclosure-policy/ | allowed | Declared policy for this page. |
| bingbotsearch | /model-card/inkling-2-small/ | allowed | Declared policy for this page. |
| bingbotsearch | /training-data-documentation/ | allowed | Declared policy for this page. |
| OAI-SearchBotsearch | / | allowed | Declared policy for this page. |
| OAI-SearchBotsearch | /legal/privacy/ | allowed | Declared policy for this page. |
| OAI-SearchBotsearch | /inkling/ | allowed | Declared policy for this page. |
| OAI-SearchBotsearch | /tinker/ | allowed | Declared policy for this page. |
| OAI-SearchBotsearch | /blog/ | allowed | Declared policy for this page. |
| OAI-SearchBotsearch | /news/ | allowed | Declared policy for this page. |
| OAI-SearchBotsearch | /legal/terms/ | allowed | Declared policy for this page. |
| OAI-SearchBotsearch | /security/disclosure-policy/ | allowed | Declared policy for this page. |
| OAI-SearchBotsearch | /model-card/inkling-2-small/ | allowed | Declared policy for this page. |
| OAI-SearchBotsearch | /training-data-documentation/ | allowed | Declared policy for this page. |
| PerplexityBotsearch | / | allowed | Declared policy for this page. |
| PerplexityBotsearch | /legal/privacy/ | allowed | Declared policy for this page. |
| PerplexityBotsearch | /inkling/ | allowed | Declared policy for this page. |
| PerplexityBotsearch | /tinker/ | allowed | Declared policy for this page. |
| PerplexityBotsearch | /blog/ | allowed | Declared policy for this page. |
| PerplexityBotsearch | /news/ | allowed | Declared policy for this page. |
| PerplexityBotsearch | /legal/terms/ | allowed | Declared policy for this page. |
| PerplexityBotsearch | /security/disclosure-policy/ | allowed | Declared policy for this page. |
| PerplexityBotsearch | /model-card/inkling-2-small/ | allowed | Declared policy for this page. |
| PerplexityBotsearch | /training-data-documentation/ | allowed | Declared policy for this page. |
| Claude-SearchBotsearch | / | allowed | Declared policy for this page. |
| Claude-SearchBotsearch | /legal/privacy/ | allowed | Declared policy for this page. |
| Claude-SearchBotsearch | /inkling/ | allowed | Declared policy for this page. |
| Claude-SearchBotsearch | /tinker/ | allowed | Declared policy for this page. |
| Claude-SearchBotsearch | /blog/ | allowed | Declared policy for this page. |
| Claude-SearchBotsearch | /news/ | allowed | Declared policy for this page. |
| Claude-SearchBotsearch | /legal/terms/ | allowed | Declared policy for this page. |
| Claude-SearchBotsearch | /security/disclosure-policy/ | allowed | Declared policy for this page. |
| Claude-SearchBotsearch | /model-card/inkling-2-small/ | allowed | Declared policy for this page. |
| Claude-SearchBotsearch | /training-data-documentation/ | allowed | Declared policy for this page. |
| GPTBottraining | / | allowed | Declared policy for this page. |
| GPTBottraining | /legal/privacy/ | allowed | Declared policy for this page. |
| GPTBottraining | /inkling/ | allowed | Declared policy for this page. |
| GPTBottraining | /tinker/ | allowed | Declared policy for this page. |
| GPTBottraining | /blog/ | allowed | Declared policy for this page. |
| GPTBottraining | /news/ | allowed | Declared policy for this page. |
| GPTBottraining | /legal/terms/ | allowed | Declared policy for this page. |
| GPTBottraining | /security/disclosure-policy/ | allowed | Declared policy for this page. |
| GPTBottraining | /model-card/inkling-2-small/ | allowed | Declared policy for this page. |
| GPTBottraining | /training-data-documentation/ | allowed | Declared policy for this page. |
| ClaudeBottraining | / | allowed | Declared policy for this page. |
| ClaudeBottraining | /legal/privacy/ | allowed | Declared policy for this page. |
| ClaudeBottraining | /inkling/ | allowed | Declared policy for this page. |
| ClaudeBottraining | /tinker/ | allowed | Declared policy for this page. |
| ClaudeBottraining | /blog/ | allowed | Declared policy for this page. |
| ClaudeBottraining | /news/ | allowed | Declared policy for this page. |
| ClaudeBottraining | /legal/terms/ | allowed | Declared policy for this page. |
| ClaudeBottraining | /security/disclosure-policy/ | allowed | Declared policy for this page. |
| ClaudeBottraining | /model-card/inkling-2-small/ | allowed | Declared policy for this page. |
| ClaudeBottraining | /training-data-documentation/ | allowed | Declared policy for this page. |
| Google-Extendedtraining and other Google AI uses | / | allowed | Declared policy for this page. |
| Google-Extendedtraining and other Google AI uses | /legal/privacy/ | allowed | Declared policy for this page. |
| Google-Extendedtraining and other Google AI uses | /inkling/ | allowed | Declared policy for this page. |
| Google-Extendedtraining and other Google AI uses | /tinker/ | allowed | Declared policy for this page. |
| Google-Extendedtraining and other Google AI uses | /blog/ | allowed | Declared policy for this page. |
| Google-Extendedtraining and other Google AI uses | /news/ | allowed | Declared policy for this page. |
| Google-Extendedtraining and other Google AI uses | /legal/terms/ | allowed | Declared policy for this page. |
| Google-Extendedtraining and other Google AI uses | /security/disclosure-policy/ | allowed | Declared policy for this page. |
| Google-Extendedtraining and other Google AI uses | /model-card/inkling-2-small/ | allowed | Declared policy for this page. |
| Google-Extendedtraining and other Google AI uses | /training-data-documentation/ | allowed | Declared policy for this page. |
| ChatGPT-Useruser-initiated; robots may not apply | / | allowed | Declared policy for this page. |
| ChatGPT-Useruser-initiated; robots may not apply | /legal/privacy/ | allowed | Declared policy for this page. |
| ChatGPT-Useruser-initiated; robots may not apply | /inkling/ | allowed | Declared policy for this page. |
| ChatGPT-Useruser-initiated; robots may not apply | /tinker/ | allowed | Declared policy for this page. |
| ChatGPT-Useruser-initiated; robots may not apply | /blog/ | allowed | Declared policy for this page. |
| ChatGPT-Useruser-initiated; robots may not apply | /news/ | allowed | Declared policy for this page. |
| ChatGPT-Useruser-initiated; robots may not apply | /legal/terms/ | allowed | Declared policy for this page. |
| ChatGPT-Useruser-initiated; robots may not apply | /security/disclosure-policy/ | allowed | Declared policy for this page. |
| ChatGPT-Useruser-initiated; robots may not apply | /model-card/inkling-2-small/ | allowed | Declared policy for this page. |
| ChatGPT-Useruser-initiated; robots may not apply | /training-data-documentation/ | allowed | Declared policy for this page. |
Advertised APIs and machine interfaces
No supported machine interfaces were advertised in the sampled HTML. An API or MCP server is optional; undiscovered interfaces remain unknown.
Public DNS and TLS endpoint
thinkingmachines.ai · observed · 104.18.1.136, 104.18.0.136, 2606:4700::6812:88, 2606:4700::6812:188
- A
- 104.18.0.136 · 104.18.1.136
- AAAA
- 2606:4700::6812:88 · 2606:4700::6812:188
- CNAME
- Unavailable for this exact hostname
- NS
- alexa.ns.cloudflare.com · kellen.ns.cloudflare.com
- MX
- 1 aspmx.l.google.com · 10 alt3.aspmx.l.google.com · 10 alt4.aspmx.l.google.com · 5 alt1.aspmx.l.google.com · 5 alt2.aspmx.l.google.com
- CAA
- {"critical":0,"type":"CAA","issuewild":"digicert.com; cansignhttpexchanges=yes"} · {"critical":0,"type":"CAA","issue":"digicert.com; cansignhttpexchanges=yes"} · {"critical":0,"type":"CAA","issuewild":"sectigo.com"} · {"critical":0,"type":"CAA","issue":"comodoca.com"} · {"critical":0,"type":"CAA","issuewild":"pki.goog; cansignhttpexchanges=yes"} · {"critical":0,"type":"CAA","issue":"ssl.com"} · {"critical":0,"type":"CAA","issue":"sectigo.com"} · {"critical":0,"type":"CAA","issuewild":"ssl.com"} · {"critical":0,"type":"CAA","issue":"pki.goog; cansignhttpexchanges=yes"} · {"critical":0,"type":"CAA","issue":"letsencrypt.org"} · {"critical":0,"type":"CAA","issuewild":"comodoca.com"} · {"critical":0,"type":"CAA","iodef":"mailto:security@thinkingmachines.ai"} · {"critical":0,"type":"CAA","issuewild":"letsencrypt.org"}
TLS: TLSv1.3
Certificate validation passed for the observed endpoint.
Issuer: Google Trust Services / WE1
Certificate valid until: Jan 27 18:08:36 2027 GMT
PageSpeed measurement
Loading the PageSpeed panel…
Growth resources
Loading growth resources…
Where to take this site next
Follow-on work suggested by the observations above.
- 01
Your top fixes list is the priority order — clear the red items before creating more content.
- 02
Re-run the combined audit after each batch of fixes
Re-run after fixes to see which checks changed; score bands are guidance, not a guarantee of ranking, traffic, or approval.
- 03
Make policy pages, publisher disclosures and substantial content reachable before applying. This audit cannot predict approval.
Apply the recommended fixes
Hand the complete evidence to a coding agent, or take the report with you.
Report notes
- 01This report covers the sampled public pages listed above. Other pages may have different results.
- 02Browser-rendered content, private backend systems and authenticated experiences may require a separate review.
- 03Unavailable observations are identified in the report. Results do not guarantee rankings, citations, security, accessibility or advertising approval.
Pages included in this report
- https://thinkingmachines.ai/
- https://thinkingmachines.ai/legal/privacy/
- https://thinkingmachines.ai/inkling/
- https://thinkingmachines.ai/tinker/
- https://thinkingmachines.ai/blog/
- https://thinkingmachines.ai/news/
- https://thinkingmachines.ai/legal/terms/
- https://thinkingmachines.ai/security/disclosure-policy/
- https://thinkingmachines.ai/model-card/inkling-2-small/
- https://thinkingmachines.ai/training-data-documentation/