Factyn
NL

How it works

Two layers, one gate.

Factyn stands between what AI writes and what you publish. It works in two layers that can be used separately or together. The Truth Layer is the core. The Style Layer is lighter and complementary.

Truth Layer

Tests every verifiable claim against external sources.

Claim extraction Factual claims, quotations, source references and causal statements are identified.
Source research Relevant primary and high-quality sources are looked up and weighed.
Verdict per claim Verified, likely correct, uncertain, likely false, or not verifiable. Each with a source reference.
High-risk detection Fabricated quotations and invented source references are handled separately. They are the most damaging errors AI makes, and the hardest for a reader to detect.

The result is not a stamp of truth. It is calibrated uncertainty — a clear picture of what stands on solid ground, and what does not.

Style Layer

Detects and softens the stylistic signatures AI leaves behind.

Typographic normalisation Em-dashes, arrows and other LLM traces are detected and replaced.
AI pattern detection Generic openers, reflexive summaries, false balance, predictable sentence structures.
Light rewriting Not to conceal AI use, but to remove artefacts that do not belong to the author's voice.

Preservation of content is strict: proper names, figures, dates and direct quotations remain untouched.

The fixer model

Present before the crisis, not after it.

Factyn is not a self-service platform. No login, no dashboard, no subscription to a tool.

Instead: a personal verification service. A trusted figure with the right instruments — who receives texts, verifies and humanises them before publication, and then steps back. Absolute discretion. A single point of contact.

Privacy and data

Confidentiality is not a precondition, but a requirement.

No training Texts are not used to train AI models. All processing runs under no-training settings.
No retention What comes in does not remain as a data trace. Zero-retention configuration where available.
Contractually anchored The guarantees sit in formal agreements, not in promises on a website.
Anonymisation In a future version, sensitive data is anonymised before processing. Verification then happens on the anonymised text; the original stays with you.

Factyn relies on leading AI models for its processing. The quality of those models is necessary for the work. The way they are used — shielded, without training, without retention — is what makes the difference.

Get in touch →