# Which YouTube channels shape AI recommendations? Google cut video citations 13.6x while timestamps and long-tail channels lead

> Across 93,598 organic YouTube citations covering 38,027 unique videos and 8,960 attributed channels (September 1 to November 20, 2026), Google AI Mode dropped YouTube's share of cited answers 13.6x from August's 53.52% peak to 3.94% in November, while ChatGPT Search stayed at 0.10%-0.27%. Inside Google AI Mode, 54.04% of Watch links point directly to exact &t= timestamps, long-tail channels outside the Top 200 supply 54.21% of attributed links, and brand-linked YouTube citations reach #1 in 33.17% of pairs (67.17% Top 3).

HTML: https://parse.gl/research/youtube-channels-that-shape-ai-recommendations · Site index: https://parse.gl/llms.txt · Published: 2026-11-20 · Updated: 2026-11-21

Which YouTube channels, URL formats, title patterns, and upload windows earn citations inside AI buying answers, and how sharply did engine reliance on YouTube shift across late 2026?

We analyzed 93,598 organic YouTube citation links across 56,334 AI answers and 22,759 buyer questions on Google AI Mode, ChatGPT Search, and Perplexity from September 1 through November 20, 2026, alongside a 1,415,352-answer July-to-November monthly trend. We normalized 51,185 URLs into 38,027 unique video IDs, attributed 8,960 channels from parsed page author metadata, measured upload age across 17,645 dated videos, and checked 56,824 brand-linked YouTube citation pairs.

## Key figures

- **distinct YouTube channels attributed:** 8,960 (61.66% of 93,598 Sep-Nov links (38,027 unique videos))
- **Google AI Mode cited answers with YouTube:** 53.52% to 3.94% (13.6x drop from Aug peak to Nov 1-20 (ChatGPT Search: 0.10%-0.27%))
- **of Watch citations deep-link to exact timestamps:** 54.04% (48,024 of 88,862 Watch links (48,023 inside Google AI Mode))
- **AI #1 and Top-3 rate on brand-linked YouTube pairs:** 33.17% / 67.17% (Across 56,824 source-level brand-linked YouTube citation pairs)

## Google AI Mode cut YouTube citations 13.6x from August's 53.52% peak to 3.94% in November

Measuring the share of cited organic answers that cite any YouTube domain across 1,415,352 cited answers from July 1 through November 20, 2026 exposes a sharp regime shift inside Google AI Mode. Google AI Mode cited YouTube on 45.65% of cited answers in July (68,397 of 149,837) and peaked at 53.52% in August (69,602 of 130,051).

From September forward, Google AI Mode pulled YouTube back sharply: 27.40% in September (34,671 of 126,520 cited answers), 12.06% in October (17,209 of 142,706), and 3.94% in November 1-20 (3,869 of 98,106), a 13.6x drop from peak. Over the exact same five months, ChatGPT Search kept YouTube between 0.10% and 0.27% of cited answers (1,452 links across September through November). In our November 1-20 Perplexity sample (1,166 cited answers), Perplexity cited YouTube on 1.72% of answers (20 answers, 69 links across 18 buyer questions).

**Share of cited answers including a YouTube domain (Jul to Nov 2026)**

| month | Google AI Mode | ChatGPT Search |
| --- | --- | --- |
| Jul 2026 | 45.65% | 0.27% |
| Aug 2026 | 53.52% | 0.18% |
| Sep 2026 | 27.4% | 0.11% |
| Oct 2026 | 12.06% | 0.14% |
| Nov 1-20 | 3.94% | 0.1% |

_Organic answers across Google AI Mode and ChatGPT Search (1,414,186 cited answers Jul-Nov 20); November covers Nov 1-20. Perplexity measured Nov 1-20 at 1.72% (20 of 1,166)._

> YouTube visibility is overwhelmingly a Google AI Mode lever, and Google reduced YouTube answer reach 13.6x between August and November.

## Watch videos supply 94.94% of links, with 54.04% deep-linked to timestamps inside Google AI Mode

Across all 93,598 YouTube citations from September 1 through November 20, 2026, standard Watch videos account for 94.94% of links (88,862 links across 49,035 distinct URLs). YouTube Shorts account for 3.49% (3,267 links across 1,506 URLs, 100% inside Google AI Mode), channel profile pages account for 1.20% (1,122 links), live streams account for 7 links, and other YouTube URLs account for 340 links.

Google AI Mode and ChatGPT Search cite fundamentally different URL structures. Inside Google AI Mode (92,077 Sep-Nov links), 48,023 Watch links (54.04% of all 88,862 Watch links; 48,024 timestamped overall) append an exact timestamp parameter (&t=), pointing buyers straight to the relevant segment inside long videos. By contrast, ChatGPT Search only recorded 1,452 YouTube links in Sep-Nov, and 62.95% of them (914 links) pointed to /@channel profile pages rather than individual video reviews or tutorials (290 Watch links).

**YouTube citation links by URL format and AI engine (Sep 1 to Nov 20, 2026)**

| URL format | Total links | Share (%) | Distinct URLs | Google AI Mode | ChatGPT Search | Perplexity |
| --- | --- | --- | --- | --- | --- | --- |
| Watch video (/watch) | 88862 | 94.94 | 49035 | 88503 | 290 | 69 |
| YouTube Shorts (/shorts) | 3267 | 3.49 | 1506 | 3267 | 0 | 0 |
| Channel profile (/@handle) | 1122 | 1.2 | 458 | 208 | 914 | 0 |
| Other YouTube URLs | 340 | 0.36 | 181 | 92 | 248 | 0 |
| Live streams | 7 | 0.01 | 5 | 7 | 0 | 0 |

_Watch links include 48,024 timestamped (&t=) links (54.04%), of which 48,023 are inside Google AI Mode. Perplexity links come from Nov 1-20 coverage._

> Structure YouTube chapters and spoken transitions around specific buyer questions so Google AI Mode can deep-link exact timestamp moments.

## 'Best/Top' roundups and '2026' titles concentrate disproportionate repeat citations

Inspecting 37,995 unique titled videos (accounting for 92,102 citation links out of 38,027 videos / 92,136 video links in Sep-Nov) reveals a clear citation multiplier for listicles and current-year framing. Videos with 'best' or 'top' in the title represent 25.61% of unique cited videos yet capture 37.53% of all citation links (1.47x their video share).

Fresh-year framing creates an even sharper concentration. Videos carrying '2026' in the title make up 18.75% of unique videos yet earn 29.12% of citation links (1.55x their video share), while videos labeled '2025' drop to 3.00% of videos and 3.17% of links.

**Share of unique cited videos vs share of citation links by title pattern**

| Item | Value |
| --- | --- |
| Best or top roundup: Unique videos | 25.61% |
| Best or top roundup: Citation links | 37.53% |
| Includes 2026 in title: Unique videos | 18.75% |
| Includes 2026 in title: Citation links | 29.12% |
| Includes 2025 in title: Unique videos | 3.00% |
| Includes 2025 in title: Citation links | 3.17% |

_Across 37,995 unique titled videos (92,102 citation links), Sep 1 to Nov 20, 2026. Categories are non-mutually exclusive._

> Current-year framing ('2026') and 'Best/Top' roundup titles attract ~1.5x more repeat citations per video than baseline cited videos.

## Half of cited videos were uploaded within 12 months (median 361 days), yet 2+ year evergreen videos hold 17%

Across 17,645 dated videos carrying 59,452 citation links (measured against November 20, 2026 in whole days), half of all cited videos were published within the past year: median upload age is 361 days (25th percentile: 164 days; 75th percentile: 755 days; mean: 609.2 days). Uploads within one year supply 50.47% of dated videos (8,906 videos) and 60.32% of citation links (35,862 links).

Citation density peaks in the 181-to-365-day window (22.78% of dated videos capturing 28.95% of links, or 17,213 links) and the 91-to-180-day window (14.38% capturing 19.70% of links). Brand-new uploads under 30 days account for just 1.64% of videos (290) and 0.99% of links (591), showing indexing lag before citations ramp up. Meanwhile, videos older than two years still account for 25.98% of dated videos (4,584) and 17.01% of links (10,111), proving high-authority tutorials remain sticky long after publication.

**Upload age distribution of cited YouTube videos (as of Nov 20, 2026)**

| Upload age bucket | Dated videos | Video share (%) | Citation links | Link share (%) |
| --- | --- | --- | --- | --- |
| Within 30 days | 290 | 1.64 | 591 | 0.99 |
| 31 to 90 days | 2058 | 11.66 | 6343 | 10.67 |
| 91 to 180 days | 2538 | 14.38 | 11715 | 19.7 |
| 181 to 365 days | 4020 | 22.78 | 17213 | 28.95 |
| 1 to 2 years | 4155 | 23.55 | 13479 | 22.67 |
| Over 2 years | 4584 | 25.98 | 10111 | 17.01 |

_17,645 dated videos accounting for 59,452 Sep-Nov citation links. Overall median: 361 days; p25: 164 days; p75: 755 days; mean: 609.2 days._

> Expect a 3-to-12-month ramp after upload for peak YouTube citation velocity, and refresh annual comparisons before existing videos age past two years.

## Specialized software and buyer-review channels top the leaderboard, led by Business Solution (1,381 links)

We mapped 61.66% of Sep-Nov YouTube links (57,715 links) to 8,960 distinct channels using parsed page author metadata, propagated across URLs sharing the same video ID. At the top of the leaderboard sit specialized buyer-guide and review channels rather than generalist entertainment networks.

Business Solution leads all channels with 1,381 links across 135 videos and 827 buyer questions, followed by SelectHub (660 links across 73 videos and 334 questions), Software Connect (650 links across 73 videos), Line45 (629 links across 71 videos), Consumer Desire (559 links from just 17 videos, averaging 32.88 links per video), Honest Outlaw (513), StartupWise (510 from 19 videos), and CRM Central (507).

**Top 12 most-cited YouTube channels in AI answers (Sep 1 to Nov 20, 2026)**

| Channel | Citation links | Unique videos | Buyer questions | Links / video |
| --- | --- | --- | --- | --- |
| Business Solution | 1381 | 135 | 827 | 10.23 |
| SelectHub | 660 | 73 | 334 | 9.04 |
| Software Connect | 650 | 73 | 271 | 8.9 |
| Line45 | 629 | 71 | 129 | 8.86 |
| Consumer Desire | 559 | 17 | 307 | 32.88 |
| Honest Outlaw | 513 | 53 | 147 | 9.68 |
| StartupWise | 510 | 19 | 340 | 26.84 |
| CRM Central | 507 | 38 | 279 | 13.34 |
| Uplatz | 465 | 124 | 295 | 3.75 |
| Be Productive | 456 | 54 | 292 | 8.44 |
| CodeLucky | 453 | 93 | 283 | 4.87 |
| GEO Mastery with Adam | 430 | 22 | 129 | 19.55 |

_57,715 attributed links mapped across 8,960 channels (61.66% of 93,598 Sep-Nov YouTube citations). Ranked by attributed citation links._

> Small, specialized software and product channels with 17 to 135 focused buyer-guide videos lead AI YouTube citations.

## 8,760 long-tail channels supply 54.21% of attributed citations, and brand-linked YouTube pairs hit #1 in 33.17%

Despite standout leaders at the top, YouTube AI citations are far from a winner-take-all concentration. Channels ranked #201 and lower (8,760 distinct channels) supply 31,285 of the 57,715 attributed links (54.21%) across 13,495 unique videos, while the Top 200 channels account for the remaining 26,430 attributed links (45.79%).

At the source-attribution level, linking cited YouTube videos to ranked brands inside the same answer yields 56,824 brand-linked YouTube citation pairs across Sep 1 to Nov 20, 2026. Across those 56,824 pairs, the linked brand ranks #1 in 33.17% of cases and finishes inside the Top 3 in 67.17% of cases.

**Channel concentration (57,715 attributed links) and brand-pair rank rates (56,824 pairs)**

| Item | Value |
| --- | --- |
| Long-tail channels (rank 201+): link share | 54.21% (31,285 links, 8,760 channels) |
| Top 200 channels: link share | 45.79% (26,430 links, 200 channels) |
| Brand-linked YouTube pairs: Top 3 AI pick | 67.17% (56,824 pairs) |
| Brand-linked YouTube pairs: #1 AI pick | 33.17% (56,824 pairs) |

_Long-tail split across 8,960 attributed channels plus source-level rank outcomes across 56,824 brand-linked YouTube citation pairs, Sep 1 to Nov 20, 2026._

> You do not need mega-channel scale to earn YouTube citations: over half of attributed links come from niche channels ranked outside the Top 200, and brand-linked YouTube citations reach Top 3 in two-thirds of pairs.

## Internet Services (30.64%), Energy (25.40%), and Apps (24.58%) rely heaviest on YouTube across 40 industries

Across all 40 Parse industry groups with at least 500 Google AI Mode answers between September 1 and November 20, 2026, YouTube citation usage varies more than 4x by vertical. Internet Services leads at 30.64% of Google AI Mode answers (4,990 answers), followed by Energy (25.40%), Apps (24.58%, 3,300 answers), Collaboration (24.50%), Consumer Electronics (24.28%), Media and Entertainment (23.19%), Artificial Intelligence (22.81%), and Software (22.04%, 16,586 answers).

At the low end sit Clothing and Apparel (8.44%), Government and Military (8.35%), and Gaming (7.23%). Across ChatGPT Search, YouTube stays below 0.4% in 39 of 40 industry groups, crossing 1% only in Sports (1.684% of 2,732 answers).

**YouTube answer citation rates across 40 Parse industry groups (Sep 1 to Nov 20, 2026)**

| Industry group | Google answers | Google YouTube (%) | ChatGPT answers | ChatGPT YouTube (%) |
| --- | --- | --- | --- | --- |
| Internet Services | 4990 | 30.64 | 5178 | 0.039 |
| Energy | 1126 | 25.4 | 1195 | 0.084 |
| Apps | 3300 | 24.58 | 3369 | 0.208 |
| Collaboration | 992 | 24.5 | 1036 | 0 |
| Consumer Electronics | 2599 | 24.28 | 2754 | 0.363 |
| Media and Entertainment | 3191 | 23.19 | 3376 | 0.355 |
| Artificial Intelligence | 5419 | 22.81 | 5701 | 0.053 |
| Administrative Services | 2423 | 22.49 | 2552 | 0 |
| Software | 16586 | 22.04 | 17002 | 0.012 |
| Privacy and Security | 516 | 21.71 | 553 | 0 |
| Sports | 2601 | 21.53 | 2732 | 1.684 |
| Design | 1104 | 19.93 | 1177 | 0 |
| Manufacturing | 2919 | 19.84 | 2991 | 0.067 |
| Hardware | 5163 | 19.43 | 5378 | 0.074 |
| Education | 3608 | 18.99 | 3793 | 0.211 |
| Advertising | 567 | 18.87 | 599 | 0.334 |
| Science and Engineering | 748 | 18.45 | 794 | 0 |
| Information Technology | 10008 | 18.24 | 10350 | 0.01 |
| Food and Beverage | 3700 | 18.11 | 3944 | 0.127 |
| Data and Analytics | 8049 | 16.44 | 8225 | 0.012 |
| Agriculture and Farming | 1359 | 16.11 | 1415 | 0.141 |
| Sustainability | 1159 | 15.62 | 1230 | 0.081 |
| Payments | 662 | 15.26 | 674 | 0 |
| Transportation | 5194 | 14.65 | 5376 | 0.223 |
| Commerce and Shopping | 8581 | 14.19 | 9020 | 0.078 |
| Events | 583 | 13.72 | 617 | 0 |
| Consumer Goods | 5492 | 13.57 | 5779 | 0.087 |
| Content and Publishing | 4113 | 13.52 | 4196 | 0.167 |
| Biotechnology | 707 | 13.15 | 750 | 0 |
| Sales and Marketing | 5701 | 13.03 | 5814 | 0.034 |
| Real Estate | 4779 | 12.97 | 5035 | 0.079 |
| Financial Services | 11343 | 12.65 | 11834 | 0.042 |
| Travel and Tourism | 2193 | 12.63 | 2334 | 0.043 |
| Blockchain and Cryptocurrency | 554 | 11.91 | 595 | 0.168 |
| Health Care | 9644 | 11.31 | 9961 | 0.02 |
| Professional Services | 5741 | 10.94 | 6048 | 0.017 |
| Community and Lifestyle | 6010 | 10.18 | 6279 | 0.111 |
| Clothing and Apparel | 1624 | 8.44 | 1706 | 0.059 |
| Government and Military | 671 | 8.35 | 713 | 0 |
| Gaming | 2574 | 7.23 | 2653 | 0.113 |

_Industry groups with >= 500 Google AI Mode answers. Shows share of organic answers citing any YouTube domain on Google AI Mode and ChatGPT Search._

> Benchmark your sector's Google AI Mode YouTube rate before sizing video budgets: B2B software and electronics pull YouTube 2x to 4x more often than apparel or gaming.

## On 20,233 matched Google AI Mode questions, YouTube doubles total citations without narrowing brand lists

To test whether YouTube citations appear alongside shorter or longer recommendations, we isolated 20,233 Google AI Mode buyer questions answered both with YouTube (51,779 answers) and without YouTube (137,019 answers) in the September-November window.

Answers citing YouTube carry nearly twice as many total citations (5.27 vs 2.73) and run 453 characters longer (3,553 vs 3,100 characters), while naming virtually the same number of brands (6.20 vs 6.33) and naming at least one brand slightly more often (95.66% vs 94.54%). YouTube arrives as added multi-source proof inside richer comparisons rather than replacing written lists.

- **Avg total citations per answer:** 5.27 vs 2.73 (With YouTube vs without YouTube (20,233 matched questions))
- **Avg answer length (characters):** 3,553 vs 3,100 (+453 characters when Google AI Mode cites YouTube)
- **Avg brands named per answer:** 6.20 vs 6.33 (Video adds depth without shrinking competitive shortlists)
- **Answers naming at least one brand:** 95.66% vs 94.54% (51,779 YouTube vs 137,019 non-YouTube answers)

> Expect YouTube citations to expand answer depth alongside written docs and reviews rather than substituting for text pages.

## What to do next: size YouTube for Google AI Mode, chapter timestamps, and long-tail channels

Focus video investment where the telemetry confirms engine retrieval. Publish structured Watch videos (94.94% of YouTube links) with explicit chapter timestamps so Google AI Mode can deep-link exact moments (54.04% of Watch citations), and refresh 'Best/Top' and current-year ('2026') titles that concentrate 1.47x to 1.55x more links per video.

Pair owned product videos with outreach to specialized niche channels (ranks 201+ account for 54.21% of attributed citations across 8,760 channels). Track monthly engine shifts separately, since Google AI Mode cut YouTube answer share 13.6x from August to November while ChatGPT Search cites /@channel profiles (62.95% of its YouTube links) more than individual video pages.

**Operator playbook by video format and AI engine**

| Strategic goal | Channel & title format | Primary AI engine | Observed benchmark |
| --- | --- | --- | --- |
| Capture timestamp deep-links | Structured Watch videos with chapter timestamps | Google AI Mode | 54.04% of Watch links carry &t= timestamps |
| Scale multi-prompt citation reach | Specialized niche channels with Best/Top & 2026 titles | Google AI Mode | 54.21% of attributed links from rank 201+; 37.53% Best/Top |
| Convert brand-linked video citations | Clear spoken & visual brand coverage in Watch videos | Google AI Mode | 33.17% #1 / 67.17% Top 3 across 56,824 brand pairs |
| Protect channel-level visibility | Clear /@handle branding and channel authority | ChatGPT Search | 62.95% of ChatGPT YT links hit /@channel |

_Grounded in 93,598 Sep-Nov YouTube citation links across Google AI Mode, ChatGPT Search, and Perplexity._

> Prioritize chaptered Watch videos, current-year buyer titles, and specialized long-tail channels while sizing spend against Google AI Mode's autumn YouTube pullback.

Related: [Plan video and content gaps in your market](https://parse.gl/use-cases/plan-content) · [Find YouTube creators and pages to pitch](https://parse.gl/use-cases/find-pages-to-pitch) · [YouTube creator partnerships for AI search](https://parse.gl/blog/youtube-creator-partnerships-for-ai-search) · [YouTube AI visibility strategy](https://parse.gl/blog/youtube-ai-visibility-strategy) · [YouTube vs Reddit in AI search (domain study)](https://parse.gl/research/youtube-most-cited-source-ai-answers) · [Query cited YouTube pages via Parse MCP](https://parse.gl/mcp)

## Study scope and counting rules

The core channel and format analysis covers 93,598 organic YouTube citation links across youtube.com, youtu.be, m.youtube.com, music.youtube.com, and kids.youtube.com from September 1 through November 20, 2026 (56,334 distinct AI answers, 22,759 buyer questions, 51,185 distinct URLs, and 38,027 unique video IDs). The five-month macro trend (July 1 to November 20, 2026) spans 1,415,352 cited organic answers across Google AI Mode, ChatGPT Search, and November 1-20 Perplexity.

Channel attribution maps 61.66% of links (57,715 links across 8,960 distinct channels) using parsed page author metadata, propagated by normalized video ID when identical videos lacked repeated author metadata. Channels ranked #201+ (8,760 channels) form the long-tail tier (31,285 links). Video upload age covers 17,645 dated videos (59,452 links) measured in whole days relative to November 20, 2026. Brand-pair rank outcomes cover 56,824 source-level brand-linked YouTube citation pairs.

Scope note: unlike Parse's YouTube vs Reddit domain study (which measures strict-public youtube.com citation share over August 23 to November 20, 2026), this report covers all organic answers across all YouTube domains from September 1 to November 20, 2026 and measures cited-answer reach alongside channel-level and source-level outcomes.

## Method

We analyzed 93,598 organic YouTube citation links across 56,334 AI answers, 22,759 buyer questions, 51,185 distinct URLs, 38,027 unique video IDs, and 8,960 attributed channels on Google AI Mode, ChatGPT Search, and Perplexity from September 1 through November 20, 2026, plus 1,415,352 cited answers across July 1 through November 20, 2026 for the five-month trend and 56,824 brand-linked YouTube citation pairs.

## Data

- [Full study dataset (CSV)](https://parse.gl/research/youtube-channels-that-shape-ai-recommendations/dataset.csv): All 81 verified rows covering Jul-Nov monthly trends, URL formats, title patterns, upload age buckets, Top 12 channels, 40 industry groups, matched-question depth, and source-level brand-pair rates.

## Sources

- [BrightEdge: How Google AI Overviews and ChatGPT use YouTube differently](https://help.brightedge.com/resources/weekly-ai-search-insights/google-ai-overviews-vs-chatgpt-youtube-usage)
- [Search Engine Land: AI search engines cite Reddit, YouTube, and LinkedIn most](https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138)
- [Ahrefs: Do AI assistants prefer to cite fresh content?](https://ahrefs.com/blog/do-ai-assistants-prefer-to-cite-fresh-content/)
- [Parse: YouTube vs Reddit in AI answers: citations and reach](https://parse.gl/research/youtube-most-cited-source-ai-answers)
- [Parse: Does AI prefer pages with the current year in the title?](https://parse.gl/research/does-ai-prefer-current-year-titles)

## For agents

- [Study dataset (CSV)](https://parse.gl/research/youtube-channels-that-shape-ai-recommendations/dataset.csv)
- Any brand's shares and rivals, with 95% ranges: `https://agent.parse.gl/v1/tools/check_brand?subject=%3Cbrand-domain%3E`
- Sample answers in a market: `https://agent.parse.gl/v1/tools/search_answers?args=%7B%22filter%22%3A%7B%22market%22%3A%5B%22%3Cmkt_id%3E%22%5D%7D%7D`
- Cited pages in a market: `https://agent.parse.gl/v1/tools/search_cited_pages?args=%7B%22filter%22%3A%7B%22market%22%3A%5B%22%3Cmkt_id%3E%22%5D%7D%7D`
- Search the index: `https://parse.gl/api/public/search?q=<words>`

Reuse: Figures and tables on this page are free to republish with attribution: a link to https://parse.gl/research/youtube-channels-that-shape-ai-recommendations beside them. A standard followed link (dofollow) is preferred.

- No account: Public figures and up to 5 examples per list, about 120 brands a day per address.
- Free account: $5 in credit, your brand's first check (20 buyer questions) and a free weekly check.
- After your first top-up: Exports and deeper search results. The first top-up starts at $5.
- Live checks: $0.01 per question per engine ($0.05 on Gemini and Claude).
- Full public answers and pages: 25 free per day, then $0.005 each.
- Public answers older than 90 days: $0.001 each the first time your workspace counts or lists one, then yours.
- An agent can save work before signup: `POST https://agent.parse.gl/v1/workspaces`
- Connect an assistant: `https://agent.parse.gl/mcp`

[Prices](https://parse.gl/pricing.md) · [Map of the site](https://parse.gl/llms.txt)
