Limited Beta

Pinpoint every AI sentence.

Submit without surprises. Know exactly which sentences in your document read as AI-generated, before anyone else sees it.

How it works

thesis_chapter3.pdf

ensembleAI: 71%

The transformer architecture fundamentally changed how we approach sequence modelling tasks.

93%

Attention mechanisms enable the model to weigh the relevance of each token dynamically.

88%

However, computational complexity grows quadratically with sequence length.

54%

My own experiments with sparse attention showed mixed results on smaller corpora.

11%

This limitation has prompted a wave of research into more efficient attention variants.

79%

The results require further validation on domain-specific benchmarks.

31%

3 AI segments

3 human segments

multi-engine analysis

Capabilities

A proprietary ensemble. One evidence-backed verdict.

Semantic Analysis

Pattern Recognition

Deep-trained pattern recognition across writing style, sentence construction, and vocabulary distribution — calibrated on academic and general-domain text.

Statistical Signals

Probability Analysis

Statistical and probability analysis that identifies the sampling signatures AI generation leaves behind at the token level.

Authorship Profiling

Stylometric Baseline

Stylometric consistency analysis that measures drift across your document and against your own historical writing baseline.

Process

From document to verdict in minutes.

Upload your document

Drop a PDF. Ensemble parses it, classifies every section, and presents an annotated Document Map for you to review.

Multiple engines analyse in parallel

Each engine analyses the selected prose independently. Results are combined by a calibrated scoring model — no shared state between engines.

Segment-level forensic report

Every 400-word window is scored independently. The ensemble aggregates them into a final verdict with supporting evidence.

Segment-level analysis

Not a single document score. Every 400-word window is scored independently, so you know which segments are flagged.

Arabic and English

Full right-to-left layout and bilingual interface. Built for international academic contexts.

Authorship fingerprinting

After 3 scans, Ensemble builds a stylometric baseline from your writing history and flags documents that deviate from it.

Ready to scan your document?

Start free. Credits top up on demand. No subscription.