Assessment built for language teaching. Every language. Every level.
XAM generates questions natively in Arabic, Urdu, French, Spanish and English, in the language of your source material, at the right CEFR level, with correct script from the first character. No translation. No broken encoding. No reformatting.
Minimum 2 seats. Or fill in the enquiry form below.
- languages, natively
- 5
- RTL repair (Arabic & Urdu)
- Auto
- CEFR calibration
- A1 to C2
- text to assessment
- < 60 sec
languages, natively
RTL repair (Arabic & Urdu)
CEFR calibration
text to assessment
Trusted by language educators
In good company, in every language.
- Northwind
- Meridian
- Lumen
- Kestrel
- Verdant
- Aster
PDF extraction has been broken for Arabic and Urdu for years. XAM fixes it.
Most PDFs encode right-to-left script incorrectly at the file level, so any tool reading them gets garbled, reversed, disconnected text. No AI can generate from that. It's a structural encoding problem, not a font one, and it affects the majority of Arabic and Urdu PDFs.
Reversed, disconnected, unreadable. No AI can generate from this.
القراءة والاستيعاب
Correct order and joining, ready to generate from.
XAM identifies the encoding errors, reconstructs the correct character sequences and restores the right-to-left structure before a single question is generated. Automatically, with no manual cleanup, no re-typing, and no workarounds, however broken the original PDF was.
Every language. Every level. More questions than time.
Language teachers assess constantly: reading, vocabulary, grammar, writing. Making materials at the right level, in the right language, for several classes at once is one of the most time-consuming parts of the job.
- Arabic and Urdu PDFs extract as scrambled, unreadable text, so AI generation fails without cleanup.
- Most tools generate only in English; Arabic or Urdu output is translated and grammatically shaky.
- Getting comprehension questions to the exact CEFR level takes expert calibration, hours per text.
- A1 and C2 learners need different questions from one syllabus, doubling the workload.
- Questions live with the teacher and disappear when they leave.
- RTL reconstruction repairs Arabic and Urdu encoding automatically, before generation, no cleanup.
- XAM generates natively in the source language. Arabic in, Arabic out, no translation step.
- CEFR calibration is built into the level and difficulty settings; pick the level and XAM matches it.
- Generate A1 and C2 versions of the same text in two runs.
- Approved questions accumulate in one shared institute bank that outlasts any teacher.
- Five languages, natively
- RTL papers, correct
- A bank that stays
Five languages. Native generation in all of them.
XAM doesn't translate from English. It processes your document in its original language and generates questions in that same language, which reads more naturally than translated output ever does.
| Language | Script direction | RTL auto-repair | Native generation |
|---|---|---|---|
| Arabic | Right to left | Yes | Yes |
| Urdu | Right to left | Yes | Yes |
| French | Left to right | Not needed | Yes |
| Spanish | Left to right | Not needed | Yes |
| English | Left to right | Not needed | Yes |
RTL auto-repair applies to Arabic and Urdu, the languages most affected by PDF encoding errors. French, Spanish and English PDFs don’t have this issue and are processed directly.
The right question for the right learner, every time.
Set the proficiency level before generating and XAM calibrates vocabulary, sentence complexity and question depth to match. The same source text produces structurally different questions at A1 and C1.
| Level | Proficiency | How XAM calibrates |
|---|---|---|
| A1 | Beginner | Simple vocabulary, short sentences, everyday topics. Primary-style audience. |
| A2 | Elementary | Common vocabulary, familiar topics, basic grammar. Short answer with scaffolding. |
| B1 | Intermediate | Wider vocabulary, unfamiliar topics, basic inference. Multiple choice with plausible distractors. |
| B2 | Upper intermediate | Abstract topics, nuanced vocabulary. Short essay and analysis. Difficulty set to Hard. |
| C1 | Advanced | Complex texts, idiomatic language, academic register. Long-form evaluation questions. |
| C2 | Mastery | Near-native complexity. Critical analysis of authentic texts, at the Create level. |
For mixed-level classes, generate two versions in two runs, each calibrated to the right level, and deliver via separate links.
Upload the Arabic PDF that used to defeat every tool.
The teachers who've fought scrambled extraction for years are the ones this was built for. Drop in the document that never worked, and the questions come back in clean, correct Arabic or Urdu, ready to review.
Every feature a language teacher needs.
The question types, delivery and export that language assessment actually calls for.
Upload any teaching text
Reading passages, grammar exercises, listening transcripts, authentic articles, past papers, publisher textbooks. XAM extracts content whatever the source language.
PDF · DOCX · TXT · all five languages
New passages from your source
Turn Creative Mode on and XAM writes fresh reading passages at the same vocabulary level, grammar and theme as your text. Ideal for unseen comprehension.
Unseen tasks, same targets
Eight question types
Multiple choice for vocabulary and grammar, short answer for comprehension, essay for writing, fill-in for grammar, matching, sorting and true/false.
Mix types in one run
Digital delivery, no LMS
Publish a shareable link; learners answer on any device with no account. Objective questions auto-grade, written responses go to the Teacher Panel.
Any device · no learner account
Print-ready papers, any language
Export branded PDF or Word papers with full layout control, including right-to-left layout for Arabic and Urdu exam papers.
RTL and LTR layout · branding
A bank that grows with the institute
Every question any teacher generates or imports joins one shared bank, filterable by language, level, type or source text. A permanent institutional resource.
Filter by language, CEFR level, type
Generate, deliver, grade. Without leaving the platform.
The whole cycle for a language assessment, from a source text to graded results, with RTL repair at the start and no LMS at any step.
- 01
Upload the source text
Any reading passage, grammar exercise or authentic text. XAM repairs RTL encoding automatically before processing.
- 02
Generate
Set the language, CEFR level, question types and Bloom's level. Up to 10 questions in under 60 seconds.
- 03
Review and approve
Every question comes to the teacher. Edit in plain language or by hand. Nothing saves without sign-off.
- 04
Export or publish
A print-ready PDF or Word paper with RTL layout for Arabic and Urdu, or a shareable link for digital submission.
Learner data stays within your institute.
GDPR-compliant
Meets GDPR for handling, storage, security and user rights. Fit for EU institutes and EU learner data.
DPA available
A Data Processing Agreement for institutes that require one. Request at privacy@xam.pro.
Encrypted storage
Source texts, questions and learner responses are encrypted at rest (AES-256) and in transit (TLS).
Learner data, contained
Responses are visible only to your teachers and admins, and never shared across institutes.
XAM never uses your teaching content, source texts, generated questions or learner responses, to train AI models.
Per-teacher pricing. A shared institute-wide bank.
From $17 per teacher each month, with a two-seat minimum. Every seat is full Educator Pro and credits pool across the institute. Annual billing saves 20%.
- Every seat is full Educator Pro, nothing held back.
- 10,000 credits a teacher, pooled across the institute.
- Add or remove seats any time. Annual billing saves 20%.
- Top-ups are $2 per 1,000 credits and never expire.
What it costs, in real terms.
A five-teacher institute is $85 a month, less than one hour of teacher time, for a tool that saves multiple hours per assessment, per teacher, every week.
Team size
≈ 50,000 questions a month, pooled across the whole team.
Any teacher can start on the Free plan first, 150 credits a month, no card, enough to generate and review a full comprehension assessment in any of the five languages with their own materials.
One institute, five languages, one shared bank.
Whether your teachers work in Arabic, Urdu, French, Spanish or English, their approved questions build one shared institute resource. New teachers inherit it; departing ones leave it behind.
What language teams say once it's in place.
I used to lose a Sunday afternoon to the end-of-term paper. I uploaded my notes on Monday morning and had it ready before first period.
[Teacher name][Subject] teacher, [School] Across the department, the papers finally look like they came from the same place. Same standard, same answer key, every time.
[Department head name]Head of Science, [Institution] The Arabic came out clean from a scanned PDF that every other tool mangled. That alone saved me an evening of retyping.
[Teacher name]Arabic teacher, [School] I set the difficulty, skimmed the questions, changed two, and sent it. It felt like having a teaching assistant who never gets tired.
[Teacher name]Maths teacher, [School] Being able to share a link and have the marking done for me changed how I run weekly quizzes. I get the class time back.
[Tutor name]Private tutor, [City] Onboarding the whole team took an afternoon, not a term. Everyone works from the same question bank now.
[Coordinator name]Assessment lead, [Institution]
Answers for the people who evaluate these things.
Yes, and it's the thing to evaluate yourself before committing. Upload one of your own Arabic or Urdu PDFs on the Free plan and check the extracted text and the questions. RTL reconstruction runs automatically before generation, and if the source has encoding errors (most do), XAM detects and repairs them. What comes back is correct, readable Arabic or Urdu.
Natively, not translated. XAM processes the source in its original language and generates questions in that same language. Translation brings grammatical errors, unnatural phrasing and vocabulary that doesn't match the register of the text. The same holds for French and Spanish.
Calibration runs through the Target Audience and Difficulty settings, which adjust vocabulary range, sentence complexity and inferential demand. It isn't perfect at every level in every language, so review before use, but A1 and C1 questions from the same text differ in the ways that matter, and the Review Panel lets you edit anything that misses your standard before you approve it.
Not currently. XAM generates written assessment from text-based sources; it doesn't produce audio or assess spoken production. For the written skills, reading comprehension, vocabulary, grammar and writing, it covers the question types language assessment needs. Speaking and listening are outside its scope for now.
Yes. Publish any assessment via a shareable link; learners open it on any device, anywhere, with no account. Objective questions auto-grade on submission and written responses appear in the Teacher Panel. For multilingual cohorts, generate a separate assessment per language and share the right link with each group.
No. On the Institutional plan, every question any teacher generates or imports feeds the institute's shared bank, accessible to all authorised staff. When a teacher leaves, their questions stay for whoever takes over the classes. The bank belongs to the institute, not the individual.
Still have a question?
Explore Help CenterTell us about your institute.
Fill in a few details and we'll be in touch within one working day, with pricing for your team size, a walkthrough of the multilingual features, or answers on Arabic, Urdu or CEFR calibration.
- Full Educator Pro on every seat
- One shared pool of credits
- Centralised institute account
- GDPR-ready, DPA on request
Ready to see it work with your own materials?
The best proof is uploading one of your own Arabic, Urdu, French or Spanish documents and seeing what XAM generates. It takes about sixty seconds.
Free plan. 150 credits a month. RTL reconstruction included.