Frequently asked
questions.

The web editor, forever: spelling correction (200,000 credits a month) with a rule explanation on every fix, word lookups, personal statistics, and 100 AI-agent actions a month. The editor also shows you how many issues the deeper categories found, so you can decide whether Pro is worth it on evidence, not promises.

Start Pro with a card on file and nothing is charged for 30 days — cancel before the month ends and you pay nothing. The card is required once per trial (each card can claim one free month). If you keep it, Pro is $10 a month billed annually ($120/year) or $25 month to month.

One action is one run of an AI agent: a paraphrase, a tashkeel pass, a summary, an expansion, an AI-detection check, a humanizer pass, or an email draft. Proofreading draws on a separate checking allowance so large on paid plans (1,000,000 credits a month) that no individual writer approaches it. Free includes 100 agent actions a month, Pro and Team include 2,000 per seat.

A shared writing standard: institutional style guides your writers apply automatically, a shared dictionary and name lexicons, team analytics (linguistic safety, time saved, improvement), centralized billing, and member management. From 3 seats at $15/seat/month billed annually.

Our newest generation of Arabic proofreading models, currently reserved for business and government customers. The public editor and self-serve plans run the proven v2 engine; v3 is deployed with your institution — cloud, sovereign cloud, or fully on-premises — and tuned to your style guide and domain. Book a demo to see it.

By organization size and deployment model (cloud, sovereign cloud, or on-premises). Government and education have dedicated tracks, and agreements can be invoiced in AED or SAR with VAT itemized. Book a demo and we will scope it with you.

No, and it is not meant to be. It supports proofreaders by catching what a person can miss and removing most of the mechanical correction work, but the editorial judgement stays human.

Paraphraser, Summarizer, Expander, Tashkeeler, Humanizer, AI Detector, Email Writer, and AI Chat, all inside the editor's AI panel. Agents that transform text work on your selection or the whole document.

No. Every agent produces a suggestion attached to the text it would change. You review it, then accept, copy, or ignore it. If the document changed while the agent was working, the suggestion is skipped rather than misapplied.

Through AI credits included in paid plans. Each request is checked against your balance before it runs, so usage is predictable for individuals and administrable for institutions.

The engine distinguishes 12 error classes, surfaced in the editor as 8 color-coded categories: spelling, grammar, morphology, semantics, style, punctuation and formatting, proper names, and institutional style guide violations. Sub-checks include quotations, dates, and numbers.

On the company benchmark sentence, Microsoft Word flags 2 errors; WriteX flags 6, each with the grammatical rule explained. Arabic morphology and semantics are simply not modeled by generic checkers.

It runs a chain of specialised AI and natural-language-processing components rather than one general model, which is what lets it classify an error, explain the rule, respond in real time, and run inside your own network.

No. It is built to support proofreaders: it catches errors a reader can miss and removes most of the mechanical correction work, while editorial judgement stays with the person.

Yes. Guides are assigned to team members, so the newsroom, legal, and social teams each write under their own rules, while admins manage everything from one screen.

Each guide starts from a national writing standard (for example Saudi Arabia's official conventions) and you override from there, so you configure only what makes your institution different.

Yes. Guides export and import as files, so a guide built in a test workspace or by a consultant can be brought into production without rebuilding it rule by rule.

The paraphraser is built to preserve meaning while changing form. You always see original and suggestion side by side, and nothing is applied until you accept.

A set of one-click transforms for institutional text: numbers formatting, terms unification, versus formatting, diacritical marking, and language switching. They reshape correct text to fit a context, rather than fixing errors.

Yes. The institutional dictionary and name lexicons let you register executives, brands, and place names so they are enforced exactly as approved, and never flagged as errors.

Yes. Arabic corrections explain in Arabic grammatical terms; English corrections explain in plain English. The interface itself is bilingual.

Yes. Explanations are written to be understood at any age or educational level, which is why schools and new learners use it as a teaching aid rather than only a proofreader.

Double-click it in the editor. A panel opens with the meaning, synonyms, and, for Arabic, the root, pattern, part of speech, and conjugations.

Yes. Dashboards export to reports for management reviews, and the same numbers are available per team and per writer.

No. Writers see their own statistics; managers see team and organization rollups according to the roles your admin configures.

From Microsoft AppSource, or directly inside Word, PowerPoint, and Outlook via the built-in add-in store. Search for Lisan, sign in, and the task pane is live.

Yes. The REST API takes text and returns classified corrections; the JavaScript SDK embeds live checking in your own web systems. Both are available on institutional plans.

Both. The engine detects the language per passage and applies the right rule set, so bilingual documents, common in Gulf institutions, are fully covered.

Not yet. Arabic and English are live in the engine and in every surface. Spanish is in development on the Correcto models and will be announced when it ships; contact us if you need it on a specific timeline.

A shipped Spanish writing assistant (web app, Chrome extension, Word add-in), purpose-built Spanish correction and NER models, tone and formality technology, and the team behind them. Spanish support in WriteX is being built on that foundation.

Yes. The engine is language-modular, Spanish is next, and the roadmap adds languages based on institutional demand.

No. The on-premises deployment keeps documents, models, and analytics inside your network. Updates are delivered as packages your team applies on your schedule.

Yes. White-labeling is available for institutional deployments; the Saudi Electricity Company runs its customized copy as Faseeh.

No. The engine is Lisan's own Relay Architecture: 32 specialised AI components developed in-house, which is why it runs on-premises, responds in real time, and explains every correction.

Yes. Sign in with your Lisan account anywhere; your style guides, dictionaries, and statistics follow you.

Office add-ins and browser extensions deploy centrally via your management tooling, and the engine itself can run on-premises. Your IT team gets a deployment guide during onboarding.

Yes. The full engine deploys on-premises, including models and analytics, and is efficient enough to run without GPU infrastructure.

Deploy the Office add-ins and browser extensions centrally, assign style guides to teams, and manage everything from the admin console. The engine scales horizontally; the SDK covers your internal systems.

Yes. The browser extensions check any web text field, and the SDK embeds the engine directly in a custom CMS with your style book attached.

Institutional education plans exist for universities and schools; contact us with your cohort size and we will quote accordingly.

Yes. Quotations from constitutional and legal texts across most Arab countries are matched against the source, and discrepancies are flagged as quotation errors.

Not on an on-premises deployment. The engine, models, and analytics all run inside your perimeter.

Institutional API and SDK plans come with an agreed throughput. Tell us your expected volume and we will size it, including a dedicated or on-premises deployment if needed.

Arabic and English today, with Spanish coming soon. The request carries the language, and the response returns classified corrections with offsets and explanations.

Technically yes, but teams consolidate quickly: one engine, one style guide, and one analytics view across both languages beats maintaining two subscriptions.

For institutional writing, yes: grammar, style, punctuation, named entities, and whole-sentence correction, plus the style-guide and analytics layers Grammarly does not have. Grammarly has more consumer polish features; WriteX has the governance.

Most institutions keep both for a term, then consolidate once the Arabic gap and the reporting become the deciding factors. We will scope a migration with your admins.

No. WriteX is a Word add-in from AppSource: a task pane beside your document with the full engine behind it.

No. They can run together; most teams turn Editor's Arabic checking off to avoid duplicate underlines and leave it on for English if they prefer.

Because for Arabic they are not the same category of tool. Editor checks spelling; WriteX analyses grammar, morphology, semantics, entities, and quotations, and enforces your institution's writing standard with reporting behind it.

Yes, and they often do: teams draft with an LLM, then run WriteX as the compliance and correctness gate before anything is published or sent.

No. It is Lisan's own Relay Architecture: 32 specialised language components chained together, which is why it responds in real time, explains each correction, and can run on your own servers.

Their language quality improves, but the structural differences do not: a generative model still rewrites rather than classifies, still cannot promise the same output twice, and still runs on infrastructure you do not control.

Yes. Text you type here is checked by the same WriteX engine that serves app.lisan.com; the suggestions and explanations are its real output, not a canned demo. The page demo is capped at 2000 characters; the full editor has no such limit.

No. The demo passes your text to the engine for checking and returns the suggestions. Nothing you type on this page is saved to a document or used to train anything.

Arabic and English are checked right here, and the language is detected automatically as you type: paste Arabic and the editor flips to right-to-left, paste English and it flips back. Spanish is coming soon, built on the Correcto models.

Hover any underline (or focus it with the keyboard) and a card opens with the category, the suggested form, and the grammatical rule behind it, exactly as in the app. Click the suggestion to apply it, or Ignore to dismiss.

Umm al-Qura, the civil calendar of Saudi Arabia, via the browser's built-in Intl engine. Religious observance based on moon sighting can differ from the civil calendar by a day.

Literally 'and nothing else': the traditional closing on cheque and contract amounts that prevents anything being appended to the figure.

The letters stay identical, but ambiguity returns: عَلِمَ and عُلِمَ both become علم. If you need the reverse, selective tashkeel that marks only ambiguous words, that is a WriteX writing tool.

No. Diacritics attach to their word under Unicode word rules, so a fully vowelled word counts exactly once, and character counts include the marks only in the character total.

Full tashkeel vowels every letter, which suits teaching material and speeches. Selective tashkeel marks only the words a reader could misread, keeping ordinary prose clean. WriteX offers both as one-click transformers.

Arabic builds words from roots and patterns, so a word can be spelled plausibly yet built wrongly, like an incorrect plural pattern. Spelling lists cannot catch that; morphological analysis can.

It is WriteX's headline quality score: the share of text free of detected errors, tracked per writer, team, and organization in the analytics dashboards.

WriteX is Lisan's AI writing assistant: an educational proofreader that checks Arabic and English across 12 error types, enforces institutional style guides, and reports language-quality analytics. It runs in a web editor, Microsoft Office, browser extensions, Google Docs and Gmail, Windows, and via API and SDK.

The web editor has a free tier with a generous spelling-check allowance, so anyone can start writing immediately. Pro is $10 a month billed annually with the first month free, teams start at $15 a seat, and enterprise plans add the v3 engine, governance, and deployment options — see pricing.

Arabic and English are live today. Arabic is the flagship, with the deepest coverage anywhere: spelling, grammar, morphology, semantics, tashkeel, and named entities. English adds sentence-level correction. Spanish is coming soon, built on the purpose-built Spanish models and team from the Correcto acquisition.

Chatbots rewrite text and cannot tell you why. WriteX classifies each error, shows the rule behind the fix, keeps your text under your control, and can run entirely on your own infrastructure. It is real-time, explainable, and auditable.

Yes. WriteX deploys to the cloud, sovereign clouds, or fully on-premises, so ministries and enterprises keep every document inside their own perimeter. The engine is efficient enough that no GPU farm is required.

Yes. The web editor is free to start: open it, paste text, and see the corrections immediately. Institutional features like style guides and analytics are demonstrated in a guided demo; book one from the contact page.

Yes. The engine detects the language per passage, applies the right rule set, and the interface itself is bilingual with full RTL support.