What is UnTXT?

Definition: UnTXT is an AI writing detector, built and run by Sataklela OÜ. You paste your text in, and what comes back is the set of passages worth reading again — each one marked AUTHENTIC or SUSPECT, with what was noticed about it a click away. Eleven detection modules do the work underneath, and developers get all eleven of them through the API, though the app itself shows two verdicts rather than asking anybody to learn eleven colours before they can read their own result.

UnTXT lives at untxt.uk as part of the CVaaS ecosystem, and because it runs in a browser there is nothing at all to install. Signing in goes through Google OAuth. The first 1,000 words are analysed free when you sign up, which is enough to form a fair opinion of it.

Core Purpose

UnTXT solves one specific problem. When somebody reads a piece of writing and suspects that parts of it were machine written, what they need to know is which parts, and why. A percentage will never tell you which sentence to rewrite. Actually that is the whole complaint with percentage scores, and it is the one thing this does answer.

Key Capabilities

Eleven module detection engine, where each module watches for one tell: SPIKE catches AI buzzwords, SYNTAX repetitive structure, FORMAL hedging language, HUMAN the genuine markers that count in your favour, TEMPLATE boilerplate phrasing, FRONTED fronted adverbials, FORENSIC style that shifts mid document, RHETORIC the AI rhetorical frames, PUFF inflated and promotional language, FORMULA the challenge and future outlook template, and LEXICAL the discourse scaffolding underneath all of it
One reading of your work rather than two competing ones, since the eleven module breakdown feeds the Academic Weapon and the Academic Weapon is what you are shown
Documents, which keeps every check you run under its own title, alongside the date, the word count, how many passages were flagged and what it cost you
PDF export, which hands the Teacher and Professional roles a formatted A4 report whenever one is wanted
CVT gifting, so that unused credit can go to another registered user rather than sit idle
Role-based interface: Student, Teacher, or Non-academic professional
Multi-currency pricing: 14 currencies supported
A note on every marked passage saying what was noticed about it, which appears when you click and goes away when you close it
Invite and Earn, where every account carries its own referral link and both sides pick up bonus CVT once the person invited finishes a first check
Confidence Guard, which lifts the threshold for the Teacher and Professional roles because a false positive costs far more when somebody is reviewing another person's work than when they are editing their own

Primary Use Cases

User	Use case
Student	Checking an essay or dissertation for AI-generated passages before submission
Teacher	Reviewing student submissions and downloading a PDF analysis report
Professional	Scanning business documents, reports, or applications for AI authorship

Who Makes UnTXT?

UnTXT is a trademark of Sataklela OÜ, registration pending, and Sataklela OÜ is who builds and runs it. It belongs to the CVaaS family of tools, which all live under untxt.uk.

Accessing UnTXT

Go to untxt.uk and sign in. Authentication runs through Google OAuth, so there is no separate account to create first. Your first 1,000 words are free, and nobody asks you for a card before you start.

Academic Integrity & Policy Compliance

Loughborough University students: the generative AI policy for assessments and UnTXT sit together without friction, because this detects rather than generates. Audit a draft with it, and you have evidence of good scholarship rather than a problem to explain. Best practice, honestly, is to acknowledge that use in your assessment statement.

Students at UCL: the university takes a supportive rather than prohibitive line on AI, and auditing your own draft fits that comfortably. UCL asks that a final submission be "substantially your own work", which is exactly the thing a check like this lets you verify before anyone else reads it. Acknowledge the use in your assessment disclosure.

Imperial College London: the requirement there is that "using AI to create assessed work" needs explicit authorisation. Auditing a draft supports that rather than working around it, since you find the machine written passages and rewrite them, and your substantial personal contribution stays intact. Acknowledge the use in your AI tools statement.

Anywhere in the Russell Group: the shared principles emphasise transparency alongside substantial personal contribution, and both are served by checking your own work. Even so, read your own institution's AI policy, because the detail really does vary from one to the next.

Educators, at any institution: reviewing submissions for AI patterns is exactly what this was built to do, and it holds up well enough for assessment integrity work.

Everywhere else: whether UnTXT is compliant depends on the AI policy your own institution has written. Broadly it supports transparent scholarly practice, because it lets a student audit a draft before submitting it.

Start for free - no payment required