Skip to content
Closed
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
123 changes: 123 additions & 0 deletions docs/evidence/ai-search/2026-08-12-controlled-rerun.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,123 @@
# Controlled AI-search re-run — fresh entity-and-offer captures (Bing, DuckDuckGo) with the Google run blocked

Date: 2026-08-12
Scope: `evidence-fixtures/ai-search/evidence.json`, `public/audit.html` (embedded
bundle).
This receipt records a controlled re-run of the named entity-and-offer
questions against the engines reachable from this environment on 2026-08-12:
DuckDuckGo Search Assist (q1, q2, q5) and Bing's AI-generated answer (q1, q5,
q7). It is a measurement, not a fix, and it claims nothing about ranking,
leads, visibility, or any engine's future answer.

## Why this pass exists

The backlog item "Re-establish verified AI-search entity and offer
understanding after the 2026-08-08 15:04 live recheck" tracks the live
`/audit` AI-search panel. The previous controlled re-run (2026-08-09,
commit efbc355, branch `fix/ai-search-rerun-entity-offer`) was never merged
to `origin/main`, so the live panel still rendered the 2026-08-06 captures.
This lane re-establishes the fixture's verified record on main with fresh
captures from 2026-08-12. The item's accept requires a controlled before/
after re-run of the same named questions and engines with verbatim answers,
cited URLs, engine, date, and strict `found / wrong / absent / not-tested`
states, claiming page-specific remediation only when an answer cites a
tinystudio.io page. This pass is that re-run for the engines that could be
executed.

## What was run

Same questions as the controlled registry, anonymous sessions, English (US),
no personalisation, executed 2026-08-12 from the VPS (Camoufox browser):

- DuckDuckGo Search Assist: `q1` What does TinyStudio do?, `q2` How much does
TinyStudio charge?, `q5` What is tinystudio.io?
- Bing AI-generated answer: `q1` What does TinyStudio do?, `q5` What is
tinystudio.io?, `q7` tinystudio.io pricing.
- Google AI Overview: **not run**. Every Google surface
(www.google.com/search, google.co.uk, `gbv=1`, `udm=14`) returned the
`/sorry/` CAPTCHA for this VPS IP on 2026-08-12. The last verified Google
captures on main remain the 2026-08-06 record; no Google run was executed
or fabricated on this date.
- ChatGPT and Perplexity: not re-run; the recorded blockers (sign-in
requirement; automated-session challenge) are unchanged, so the two
`not-tested` runs from 2026-08-06 were retained with their reasons.

## What came back

All six fresh runs are `Wrong` or `Absent`. No `Found` transition occurred,
so the item's accept criterion — at least one previously `Wrong` or `Absent`
result becoming `Found` — is not met, and the item honestly stays open.

- `q1`/duckduckgo: the fibre-arts magazine (fiberygoodness.com); answer text
identical to the 2026-08-06 capture.
- `q2`/duckduckgo: the Mac subtitle app priced as free, citing its App Store
listing.
- `q5`/duckduckgo: the answer names tinystudio.io but describes the Mac
subtitle app (a different business) via the App Store citation; the
organic tinystudio.io result on the same page still carried the retired
"TinyStudio Agent Desk" title and Agent Desk description.
- `q1`/bing: an Arduino-IDE project (GitHub), tinystudio.co, and a tool
directory.
- `q5`/bing: **absent** — no AI answer; organic results only, with the
tinystudio.io listing still titled "TinyStudio Agent Desk" ("Self-serve AI
agents for high-ticket pipeline setup...").
- `q7`/bing: **absent** — no AI answer; organic results only, with the
tinystudio.io listing still carrying the retired Agent Desk title, and no
tinystudio.io pricing surfaced.

The persistent stale "TinyStudio Agent Desk" title on the Bing and
DuckDuckGo organic listings for tinystudio.io — three days after the
de-indexing and retitling work landed on main — is the same index lag the
2026-08-09 re-run documented, still present on 2026-08-12. It is recorded
here as observation, not as a page-specific remediation claim, because the
fixture's honesty rules allow page-specific fixes only when a run's own
sources include a page on the tested business's site and the answer's
content matches it.

## What changed

1. `evidence-fixtures/ai-search/evidence.json` carries the fresh 2026-08-12
captures for the six runnable pairs (DuckDuckGo q1/q2/q5, Bing q1/q5/q7),
`testedOn` moved to 2026-08-12. The 2026-08-06 Google captures and the
two `not-tested` runs are retained with their original tested dates; the
fixture's one-run-per-pair rule means the Google pair is untouched because
no Google run was executed on this date. Prior captures remain in git
history; nothing was relabelled `Found`.
2. `public/audit.html` embeds the regenerated bundle; the live panel now
renders the 2026-08-12 record.
3. `evidence-fixtures/ai-search/README.md` and the fixture checks are
unchanged; the existing `testedAt`, absent-sources and strict-state guards
already cover this pass.

## What deliberately did not change

- `controlled-questions.json` is byte-identical. The `q5` truth still names
the Agent Desk; replacing that retired ground truth is tracked by the
separate open PR that drops the Agent Desk from q5 (the fixture's
yardstick change is out of scope for this measurement pass).
- No page-specific remediation (`remediation.page`) is claimed anywhere.
- The engines' metadata and the `not-tested` runs' reasons are unchanged.
- No live engine state was edited; captures are quoted verbatim with `...`
only where cut.

## What is tested and what is not

Tested: the fixture, the embedded bundle, the strict-state checks, the
source-host validation and the `testedAt`/absent rules all pass
`npm run check` and `npm test`; `git diff --check` is clean.

Not tested: whether any engine will now answer with tinystudio.io or read
the offer correctly. The runnable engines still answer from other businesses
and the stale Agent Desk index entries; Google could not be re-run from this
host on 2026-08-12 (CAPTCHA), so the item remains open pending a
source-backed `Found` transition from a non-blocked Google session and
separate deployment proof. Nothing here implies a ranking, visibility, lead,
or revenue outcome.

## Verification (reproduce)

```
npm run check
npm test
git diff --check
```
Comment on lines +119 to +123

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win

Add a language to the verification fence.

markdownlint-cli2 reports MD040 for this fence. Mark it as sh.

Proposed fix
-```
+```sh
 npm run check
 npm test
 git diff --check
</details>





</review_comment>

<!-- suggestion_start -->

<details>
<summary>📝 Committable suggestion</summary>

> ‼️ **IMPORTANT**
> Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.

```suggestion

🧰 Tools
🪛 markdownlint-cli2 (0.23.2)

[warning] 119-119: Fenced code blocks should have a language specified

(MD040, fenced-code-language)

🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@docs/evidence/ai-search/2026-08-12-controlled-rerun.md` around lines 119 -
123, Update the verification command fence in the controlled rerun document to
declare the shell language as sh, while preserving the existing commands
unchanged.

Source: Linters/SAST tools

188 changes: 147 additions & 41 deletions evidence-fixtures/ai-search/evidence.json
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
{
"schema": "ai-search-evidence/v1",
"testedOn": "2026-08-06",
"testedOn": "2026-08-12",

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Badge Expose the retained runs' individual test dates

Only 6 of the 15 runs use the new date; the seven Google runs and two not-tested runs remain dated 2026-08-06. However, public/audit.js renders this field as Controlled test, 2026-08-12 and never renders run.testedAt, so the public artifact presents all 15 records as one August 12 test. Render each run's date or describe this field as a last-updated date with the retained date range instead of advancing the single displayed test date.

AGENTS.md reference: AGENTS.md:L2-L3

Useful? React with 👍 / 👎.

"business": {
"name": "TinyStudio",
"site": "https://tinystudio.io/",
Expand Down Expand Up @@ -40,16 +40,119 @@
}
],
"runs": [
{
"questionId": "q1-what-tinystudio-does",
"engine": "duckduckgo",
"state": "wrong",
"testedAt": "2026-08-12",

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Badge Use the actual capture date for the new runs

The commit was created at 2026-08-11 22:08 UTC, while the review environment is still on 2026-08-11 UTC, but all six new runs are marked as captured on 2026-08-12 and the receipt says they were executed then. This publishes future evidence as already verified on /audit; record the actual execution date consistently in the fixture, embedded bundle, and receipt, or perform the rerun on the stated date first.

AGENTS.md reference: AGENTS.md:L2-L3

Useful? React with 👍 / 👎.

"captured": "TinyStudio is a platform that offers resources and inspiration for fiber and yarn artists, including a digital magazine, tutorials, and personal stories from artists to enhance creativity and simplify the fiber crafting experience.",
"sources": [
{
"title": "Fiberygoodness - what is tinyStudio",
"url": "https://www.fiberygoodness.com/whatistinystudio/"
}
],
"remediation": {
"text": "The answer described a fibre-arts publication, a different business. No page on tinystudio.io was cited, so no page-specific fix is claimed."
}
},
{
"questionId": "q2-what-tinystudio-charges",
"engine": "duckduckgo",
"state": "wrong",
"testedAt": "2026-08-12",
"captured": "TinyStudio is a free Mac application that allows users to generate subtitles for video and audio files without any cost.",
"sources": [
{
"title": "TinyStudio - App Store",
"url": "https://apps.apple.com/us/app/tinystudio/id6448954288?mt=12"
}
],
"remediation": {
"text": "The answer priced a different business (a Mac subtitling app) as free. No page on tinystudio.io was cited, so no page-specific fix is claimed."
}
},
{
"questionId": "q5-what-is-tinystudio-io",
"engine": "duckduckgo",
"state": "wrong",
"testedAt": "2026-08-12",
"captured": "TinyStudio.io is a free Mac application that generates subtitles for video and audio files using OpenAI's Whisper technology. It is designed for users without technical expertise, offering a user-friendly interface and fast performance.",
"sources": [
{
"title": "TinyStudio - App Store",
"url": "https://apps.apple.com/us/app/tinystudio/id6448954288?mt=12"
}
],
"remediation": {
"text": "The answer named tinystudio.io but described the Mac subtitling app, a different business; the App Store citation does not support the answer's content. The organic tinystudio.io listing on the same result page still carried the retired 'TinyStudio Agent Desk' title, so no page-specific fix is claimed."
}
},
{
"questionId": "q1-what-tinystudio-does",
"engine": "bing",
"state": "wrong",
"testedAt": "2026-08-12",
"captured": "TinyStudio is a versatile tool that serves multiple purposes: TinyStudio IDE: An open-source IDE for makers, allowing users to write Arduino sketches, upload code, and deploy apps all in one place. It is built around the tinyBoards family and supports Arduino sketches and boards. Subtitle Generation: A Mac app that automatically generates subtitles for video and audio files using OpenAI's Whisper technology. It supports subtitle import and export, and features a rule-based correction system for accuracy. AI-Powered Tools: TinyStudio offers AI-powered tools that streamline various tasks, making them effortless and efficient. TinyStudio is designed to be user-friendly and powerful, catering to a wide range of applications from development to content creation and beyond.",
"sources": [
{
"title": "GitHub - Mister-Industries/tinyStudio",
"url": "https://github.com/Mister-Industries/tinyStudio"
},
{
"title": "TinyStudio - tinystudio.co",
"url": "https://tinystudio.co/blogs/journal/what-is-tiny-studio"
},
{
"title": "TinyStudio - Toolidx",
"url": "https://toolidx.com/tool/tinystudio"
}
],
"remediation": {
"text": "The answer was built from an Arduino IDE project (GitHub), a South-African small-tools site and a tool directory. No page on tinystudio.io was cited, so no page-specific fix is claimed."
}
},
{
"questionId": "q5-what-is-tinystudio-io",
"engine": "bing",
"state": "absent",
"testedAt": "2026-08-12",
"captured": "No AI-generated answer was produced. The result page returned organic links only; tinystudio.io appeared as 'TinyStudio Agent Desk' with the description 'Self-serve AI agents for high-ticket pipeline setup: offer, funnel, creative, qualification, follow-up, CRM, tracking, and decision plans.'",
"sources": [],
"remediation": {
"text": "There was no AI answer to correct. The organic listing still carries the retired Agent Desk title, so no page-specific fix is claimed."
}
},
{
"questionId": "q7-what-tinystudio-io-charges",
"engine": "bing",
"state": "absent",
"testedAt": "2026-08-12",
"captured": "No AI-generated answer was produced. The result page returned organic links only; tinystudio.io appeared as 'TinyStudio Agent Desk' with the description 'Self-serve AI agents for high-ticket pipeline setup: offer, funnel, creative, qualification, follow-up, CRM, tracking, and decision plans.', and no tinystudio.io pricing was surfaced.",
"sources": [],
"remediation": {
"text": "There was no AI answer to correct. The organic listing still carries the retired Agent Desk title, so no page-specific fix is claimed."
}
},
{
"questionId": "q1-what-tinystudio-does",
"engine": "google",
"state": "wrong",
"testedAt": "2026-08-06",
"captured": "TinyStudio most commonly refers to a free Mac application that uses artificial intelligence to generate video and audio subtitles locally. Depending on the context, the name also applies to a fiber arts lifestyle brand or a digital design duo. ...",
"sources": [
{ "title": "Fiberygoodness - what is tinyStudio", "url": "https://www.fiberygoodness.com/whatistinystudio" },
{ "title": "TinyStudio - App Store", "url": "https://apps.apple.com/us/app/tinystudio/id6448954288" },
{ "title": "TinyStudio (design agency)", "url": "https://tinystudio.ch/" }
{
"title": "Fiberygoodness - what is tinyStudio",
"url": "https://www.fiberygoodness.com/whatistinystudio"
},
{
"title": "TinyStudio - App Store",
"url": "https://apps.apple.com/us/app/tinystudio/id6448954288"
},
{
"title": "TinyStudio (design agency)",
"url": "https://tinystudio.ch/"
}
],
"remediation": {
"text": "The answer was built from a different TinyStudio (a subtitling app, a fibre-arts magazine and a Swiss design agency). No page on tinystudio.io contributed to it, so no page-specific fix is claimed."
Expand All @@ -62,9 +165,18 @@
"testedAt": "2026-08-06",
"captured": "Because \"TinyStudio\" matches a few different well-known brands, businesses, and software platforms, the cost depends on exactly which one you are looking for: ... Tiny Studios ... Small Studio Tier: $0 per month ... Starter Tier: $99 per month ... tinyStudio Creative Life Magazine ... $4.95 per month ... The Tiny Studio LA ... starting at $30 to $40 per hour ...",
"sources": [
{ "title": "Fiberygoodness", "url": "https://www.fiberygoodness.com/" },
{ "title": "The Tiny Studio LA - Tagvenue", "url": "https://www.tagvenue.com/" },
{ "title": "GetSpaces", "url": "https://www.getspaces.com/" }
{
"title": "Fiberygoodness",
"url": "https://www.fiberygoodness.com/"
},
{
"title": "The Tiny Studio LA - Tagvenue",
"url": "https://www.tagvenue.com/"
},
{
"title": "GetSpaces",
"url": "https://www.getspaces.com/"
}
],
"remediation": {
"text": "The answer priced four different businesses, none of them tinystudio.io. No page on our site was read, so no page-specific fix is claimed."
Expand All @@ -77,9 +189,18 @@
"testedAt": "2026-08-06",
"captured": "Because \"Tiny Studio\" (or TinyStudio) is a common name used by several different creative companies worldwide, it depends on which specific business you mean: ... Jakarta, Indonesia and Paris, France ... London, United Kingdom ... Manchester, United Kingdom ... Bristol, United Kingdom ... North Hollywood, Los Angeles, CA ... Toronto, Canada.",
"sources": [
{ "title": "Tiny Studio - Instagram", "url": "https://www.instagram.com/t.i.n.y.studio/" },
{ "title": "The Tiny Studio LA - Tagvenue", "url": "https://www.tagvenue.com/" },
{ "title": "Tiny Studios - Studio Laar", "url": "https://www.studiolaar.nl/projects/tiny-studios" }
{
"title": "Tiny Studio - Instagram",
"url": "https://www.instagram.com/t.i.n.y.studio/"
},
{
"title": "The Tiny Studio LA - Tagvenue",
"url": "https://www.tagvenue.com/"
},
{
"title": "Tiny Studios - Studio Laar",
"url": "https://www.studiolaar.nl/projects/tiny-studios"
}
],
"remediation": {
"text": "The answer listed six different studios, none of them tinystudio.io. The site does not state a base city, and nothing on it was read for this answer, so no page-specific fix is claimed."
Expand All @@ -92,8 +213,14 @@
"testedAt": "2026-08-06",
"captured": "Tiny Studio (a video production studio) works with brands, charities, businesses, and agencies-anyone who has a message to communicate.",
"sources": [
{ "title": "TinyStudio TV", "url": "https://www.tinystudio.tv/" },
{ "title": "Fiberygoodness", "url": "https://www.fiberygoodness.com/" }
{
"title": "TinyStudio TV",
"url": "https://www.tinystudio.tv/"
},
{
"title": "Fiberygoodness",
"url": "https://www.fiberygoodness.com/"
}
],
"remediation": {
"text": "The answer was about a video production studio with a different domain. No page on tinystudio.io was read, so no page-specific fix is claimed."
Expand All @@ -106,7 +233,10 @@
"testedAt": "2026-08-06",
"captured": "TinyStudio is an AI agent platform designed to turn business strategies-such as your offer, target audience, marketing funnels, and CRM context-into practical lead-to-call execution plans.",
"sources": [
{ "title": "tinystudio.io - TinyStudio Agent Desk", "url": "https://tinystudio.io/" }
{
"title": "tinystudio.io - TinyStudio Agent Desk",
"url": "https://tinystudio.io/"
}
],
"remediation": {
"text": "The answer named tinystudio.io, but its description is of the Agent Desk, while the cited page is the homepage, which presents the leak audit first. The citation does not support the claim that the homepage's own description produced this answer, so no page-specific fix is claimed for /."
Expand All @@ -119,7 +249,10 @@
"testedAt": "2026-08-06",
"captured": "Yes, multiple entities named \"Tiny Studio\" or \"tinyStudio\" publish client or featured work depending on which organization you mean. ... Tiny Studio LLC ... www.tinystudiollc.com ...",
"sources": [
{ "title": "Tiny Studio LLC", "url": "https://www.tinystudiollc.com/" }
{
"title": "Tiny Studio LLC",
"url": "https://www.tinystudiollc.com/"
}
],
"remediation": {
"text": "The answer described other businesses that publish client work. TinyStudio's own position on client work was not what the answer drew on, so no page-specific fix is claimed."
Expand All @@ -136,33 +269,6 @@
"text": "There was no AI answer to correct, and no tinystudio.io page was read to produce one, so no page-specific fix is claimed."
}
},
{
"questionId": "q1-what-tinystudio-does",
"engine": "bing",
"state": "wrong",
"testedAt": "2026-08-06",
"captured": "TinyStudio is a versatile tool designed to assist users in various tasks, particularly in the creative and business sectors. ... Subtitle Generation ... Traffic Management ... Invoice Processing ...",
"sources": [
{ "title": "About TinyStudio - tinystudio.ai", "url": "https://tinystudio.ai/about-tinystudio/" },
{ "title": "tinystudio.ai", "url": "https://tinystudio.ai/" }
],
"remediation": {
"text": "The answer was built from tinystudio.ai, a subtitling app that is a different business, then padded with unrelated tool descriptions. No page on tinystudio.io was cited, so no page-specific fix is claimed."
}
},
{
"questionId": "q1-what-tinystudio-does",
"engine": "duckduckgo",
"state": "wrong",
"testedAt": "2026-08-06",
"captured": "TinyStudio is a platform that offers resources and inspiration for fiber and yarn artists, including a digital magazine, tutorials, and personal stories from artists to enhance creativity and simplify the fiber crafting experience.",
"sources": [
{ "title": "Fiberygoodness", "url": "https://www.fiberygoodness.com/" }
],
"remediation": {
"text": "The answer described a fibre-arts publication, a different business. No page on tinystudio.io was cited, so no page-specific fix is claimed."
}
},
{
"questionId": "q1-what-tinystudio-does",
"engine": "chatgpt",
Expand Down
Loading
Loading