RankDude

Test log

Sapling test data

Raw results for the Sapling review. The verdict is in the review.

Tested
6 November 2026
Tester
Miles Foster
Account
Free demo account, signed up by RankDude, nothing paid to the brand
Tests
8
Screenshots
10
Environment
Google Chrome on desktop

Results

IDTestKey measurementScore
T01Accuracy on AI-Written TextAI passages caught: 5 of 55/5
T02False Positives on Human TextHuman passages flagged as AI: 7 of 301/5
T03Mixed & Lightly-Edited TextMixed passages flagged with the AI part marked: 2 of 25/5
T04Non-English TextPassages correct: 4 of 45/5
T05Document & File SupportWays in that worked: 8 of 104/5
T06Score Clarity & EvidencePoints met: 8 of 104/5
T07Scan SpeedMedian per scan: 0.5 s5/5
T08Pricing & Free-Tier LimitsFree check limit: 2,000 characters3/5
Overall (in the review)4.6/5
T01

Accuracy on AI-Written Text

5/5Tested
Method
Five AI-written passages of 78 to 91 words (remote work, the Industrial Revolution, sleep, a cover letter, electric cars) were pasted once each; the first score counted, with 50 percent AI as the line.
AI passages caught
5 of 5
Lowest AI score
98.7% AI
Remark
Scale 5: 47 to 50 of 50 caught. 5 of 5 is 100 percent, so the grade is 5; the sample is five passages, not fifty.

Sapling: An AI-written passage about the Industrial Revolution scored 99.9% AI, with every sentence highlighted red.

Sapling: An AI-written passage about the Industrial Revolution scored 99.9% AI, with every sentence highlighted red.
T01-1 An AI-written passage: 99.9% AI, every sentence highlighted.
Raw notes

Five AI-written passages of 78 to 91 words (remote work, the Industrial Revolution, sleep, a cover letter, electric cars) were pasted once each; the first score counted, with 50 percent AI as the line.

All five scored between 98.7% and 99.9% AI, and every sentence was highlighted red.

The Spanish and German AI passages also scored 99.9% AI (see the non-English test).

  • Five passages from one model (Claude), at default settings, not 50 from six models.
T02

False Positives on Human Text

1/5Tested
Method
Thirty human passages were pasted once each: five from Pride and Prejudice, Walden, On the Origin of Species, Moby-Dick and the Autobiography of Benjamin Franklin, and twenty-five more from books and essays published between 1776 and 1899, one passage each (Frankenstein, Dracula, Jane Eyre, Alice's Adventures in Wonderland and others).
Human passages flagged as AI
7 of 30
Rate
23%
Highest wrong score
95.4% AI
Remark
Scale 1: more than 10 of 50 flagged (over 20%). 7 of 30 is 23 percent. The method also scores 2 or lower on this test for any false-positive rate above 10 percent.

Sapling: A passage from Frankenstein (1818) scored 95.0% AI, with every sentence highlighted red.

Sapling: A passage from Frankenstein (1818) scored 95.0% AI, with every sentence highlighted red.
T02-1 A passage from Frankenstein (1818), written 200 years before any public AI: 95.0% AI.

Sapling: A passage from Walden (1854) scored 76.8% AI, with every sentence highlighted red.

Sapling: A passage from Walden (1854) scored 76.8% AI, with every sentence highlighted red.
T02-2 A passage from Walden (1854): 76.8% AI.

Sapling: A passage from Moby-Dick scored 5.3% AI with no sentence highlighted.

Sapling: A passage from Moby-Dick scored 5.3% AI with no sentence highlighted.
T02-3 A passage from Moby-Dick (1851): 5.3% AI, nothing highlighted.
Raw notes

Thirty human passages were pasted once each: five from Pride and Prejudice, Walden, On the Origin of Species, Moby-Dick and the Autobiography of Benjamin Franklin, and twenty-five more from books and essays published between 1776 and 1899, one passage each (Frankenstein, Dracula, Jane Eyre, Alice's Adventures in Wonderland and others). Every one was written long before any public AI, so any AI flag is a false positive.

Seven of the thirty scored 50% AI or more: Frankenstein 95.0%, Alice's Adventures in Wonderland 95.4%, Huckleberry Finn 91.7%, Jane Eyre 83.8%, Treasure Island 77.5%, Walden 76.8% and The Picture of Dorian Gray 62.6%.

The other 23 scored between 0% and 10.2% AI. The two non-English human passages (Don Quijote, Les Miserables) scored 0.2% and 0.4%.

Six of the seven flagged passages are from novels.

  • Thirty passages, not fifty. All are old books, so the result says nothing about student essays. The first five passages gave one flag (Walden); the other six flags came from the later twenty-five.
T03

Mixed & Lightly-Edited Text

5/5Tested
Method
Two passages each joined a human paragraph (from Great Expectations and A Tale of Two Cities) to an AI paragraph, in both orders, and were scanned once.
Mixed passages flagged with the AI part marked
2 of 2
Overall scores
2.9% and 2.4% AI
Remark
Counted as flagged because the verdict line said mixed and the AI sentences were marked. Scale 5 is 17 of 20 with accurate highlights.

Sapling: A mixed passage scored 2.9% AI overall; the sentence list shows the two AI sentences at 99.8% and 99.9% AI in red and the human ones at 0.0% in green.

Sapling: A mixed passage scored 2.9% AI overall; the sentence list shows the two AI sentences at 99.8% and 99.9% AI in red and the human ones at 0.0% in green.
T03-1 A mixed passage: 2.9% overall, but the two AI sentences score 99.8% and 99.9% AI.
Raw notes

Two passages each joined a human paragraph (from Great Expectations and A Tale of Two Cities) to an AI paragraph, in both orders, and were scanned once.

Both scored low overall, 2.9% and 2.4% AI, but both showed the line "Mixed: one passage is likely AI-generated."

The highlights and the sentence list landed on the AI part: the two AI sentences scored 99.8% and 99.9% AI and every human sentence 0.0% AI, in both passages. I ran both passages a second time in the dashboard to read the sentence scores, and the overall scores came out the same, 2.9% and 2.4%.

  • Two passages, not twenty. The AI-draft-with-human-edits kind was not run. The headline percentage alone would read as human.
T04

Non-English Text

5/5Tested
Method
Four passages were scanned once: an AI passage in Spanish, an AI passage in German, a human Spanish passage (Don Quijote) and a human French passage (Les Miserables).
Passages correct
4 of 4
Human passages flagged
0
Remark
Scale 5: 22 to 24 of 24 correct with at most one human passage flagged. 4 of 4 is 100 percent.

Sapling: A German passage written by an AI, scored 99.9% AI with every sentence highlighted red and the note Detected German.

Sapling: A German passage written by an AI, scored 99.9% AI with every sentence highlighted red and the note Detected German.
T04-1 A German AI passage: 99.9% AI, detected as German.
Raw notes

Four passages were scanned once: an AI passage in Spanish, an AI passage in German, a human Spanish passage (Don Quijote) and a human French passage (Les Miserables).

Both AI passages scored 99.9% AI. The human Spanish passage scored 0.2% and the human French passage 0.4%.

The page named the language for each ("Detected German: scored with Sapling's multilingual model").

  • Four passages, three languages. Chinese was not run. The English human set had a 23 percent false-positive rate, so four clean passages are a weak guide.
T05

Document & File Support

4/5Tested
Method
Each way in was tried on the Pro trial: paste, a PDF, a Word file, a text file, a batch of files, a web page address and a 2,190-word text, then the export, extension and API points were checked.
Ways in that worked
8 of 10
Remark
Scale 4: 8 or 9 of 10 points. The long-text point passed on the Pro trial (100,000 characters a check); the free plan stops at 2,000 characters.

Sapling: The Files tab after a batch of six files was checked, each listed with its score and verdict.

Sapling: The Files tab after a batch of six files was checked, each listed with its score and verdict.
T05-1 A batch of six files (two PDFs, two Word files, two text files), each with a score.

Sapling: The Web page tab after a Wikipedia article was checked, showing 4.0% AI and 13,349 words analysed.

Sapling: The Web page tab after a Wikipedia article was checked, showing 4.0% AI and 13,349 words analysed.
T05-2 A web page checked by address: 13,349 words, 4.0% AI.
Raw notes

Each way in was tried on the Pro trial: paste, a PDF, a Word file, a text file, a batch of files, a web page address and a 2,190-word text, then the export, extension and API points were checked.

Worked: paste; PDF; DOCX; TXT; a batch of six files (two PDFs, two DOCX, two TXT) checked together in 2.0 seconds with every verdict right; a web page address (a Wikipedia article, 13,349 words); a 2,190-word text in one check (0.6% AI, 2.1 seconds); and Export CSV, which downloaded a file with the score, verdict and a report link per file. That is 8 of 10.

Not counted: the Chrome extension (offered, not installed) and the API (it is metered and billed separately, and I did not create a key).

  • The 10-file limit and 5 MB per file are the limits the page states; I did not test them. The Wikipedia article scored 4.0% AI overall but the page said 38 passages of its 1,528 sentences were likely AI.
T06

Score Clarity & Evidence

4/5Tested
Method
Each of the ten points was checked on the scans already run, with one passage scanned three times for repeatability.
Points met
8 of 10
Remark
Scale 4: 8 or 9 of 10 points.

Sapling: A 2,190-word text scored 0.6% AI in the dashboard, with the Copy report, Share certificate and Share link buttons and the history on the right.

Sapling: A 2,190-word text scored 0.6% AI in the dashboard, with the Copy report, Share certificate and Share link buttons and the history on the right.
T06-1 A 2,190-word text in one check, with the report buttons and the saved history.
Raw notes

Each of the ten points was checked on the scans already run, with one passage scanned three times for repeatability.

Met: a plain call ("Unlikely AI-authored; appears human-written." and "Highly likely composed by AI." in the dashboard); a percentage; sentence highlighting; a shareable report (a link that expires after 3 days, a Share certificate and a Share link); a saved history (up to 5,000 rows); repeatability (the same passage scored 99.9% AI all three times); a stated limit ("False positives and negatives may occur."); and no upsell blocking the result. That is 8 of 10.

Missed: a stated margin or confidence band, and a written reason for the score. The score is one number labelled "model estimate".

  • The public detector page shows the score and a mixed-text line; the plain verdict sentence and the saved history are in the signed-in dashboard.
T07

Scan Speed

5/5Tested
Method
All 42 timed scans were timed from pressing Check until the button changed back from Checking and the score was shown, checking twenty times a second.
Median per scan
0.5 s
Slowest scan
1.7 s
Remark
Scale 5: a median under 5 seconds.
Raw notes

All 42 timed scans were timed from pressing Check until the button changed back from Checking and the score was shown, checking twenty times a second.

Median 0.5 seconds, fastest 0.3, slowest 1.7. No scan failed or queued. A batch of six files took 2.0 seconds and a 2,190-word text 2.1 seconds.

  • Passages of 73 to 140 words on the Pro trial; the free plan's speed was not measured.
T08

Pricing & Free-Tier Limits

3/5Tested
Method
The plans page was read on the test date in both billing views, and the Pro trial was used for the whole run.
Free check limit
2,000 characters
Pro, billed monthly
$25
Remark
Scale 3: a tight free tier, and a paid plan at $25 a month, in the $20 to $29.99 band; billed yearly it is $12.

Sapling: The four plans, Free, Pro, API and Enterprise, on monthly billing.

Sapling: The four plans, Free, Pro, API and Enterprise, on monthly billing.
T08-1 The plans page on 12 November 2026, monthly billing.
Raw notes

The plans page was read on the test date in both billing views, and the Pro trial was used for the whole run.

Free: $0, AI detector checks up to 2,000 characters a check (about 330 words), no card. I did not run the free plan, so I could not check a daily limit.

Pro: $25 a month, or $12 a month billed yearly, with checks up to 100,000 characters and 10 files per batch. The page lists no monthly word limit.

The API is metered with a one-time $5 minimum and Enterprise starts at 10 seats at $15 a seat a month. A new account starts a Pro trial: the account page said 30 days left, and I entered no card.

  • No plan was bought and the checkout was not opened. The cost of 1,000 words cannot be worked out because the Pro plan lists no word metering.

Coverage

Tested

  • 42 timed scans: 5 AI passages, 5 and 25 human passages from books published before 1900, 2 mixed passages, 4 non-English passages and a repeat
  • Time from pressing Check to the score on screen, for every scan
  • A batch of six files (PDF, DOCX, TXT), Export CSV, a web page check and a 2,190-word text
  • The sentence list, the share link, the saved history and the repeat of one passage three times
  • The plans page read on the test date
  • The Trustpilot page read in the same browser

Not tested or limited

  • Passages are short, 73 to 140 words, not the 300 words and 50 passages of each kind in the full method
  • Resistance to humanisers and AI model coverage were not run: no humaniser account was bought and every AI passage was written by one model (Claude)
  • Every human passage is from a book published before 1900, a kind of text I chose because no AI can have written it
  • Chinese was left out of the non-English test
  • Everything ran on the Pro trial; the free plan's 2,000-character limit and its daily behaviour were not run
  • The Chrome extension was not installed and the API was not used

Still to test

  • The other 45 AI passages from six models
  • Resistance to humanisers and paraphrasers
  • A modern human set: essays, emails and posts
  • The free plan's limits over a day
  • The API and the Chrome extension
What these results meanRead the full Sapling review