AI Features
Use AI to generate lessons, exercises, translations, and get writing assistance.
The Lex Editor integrates AI tools for lesson creation, exercise generation, translation, and content analysis.
Overview of AI Tools
| Tool | What It Does | How to Access |
|---|---|---|
| AI Lesson Builder | Shapes a Lesson Plan and writes its ordered Steps | Canvas view in the editor |
| Ask AI | Inline AI assistant for writing and editing help | Toolbar button or /askai |
| Exercise Generation | Creates exercises from lesson context | Exercise block settings |
| Translation Panel | Translates selected text between languages, with quick chips for the lesson's relevant language pairs | Select text > Translate button |
AI Lesson Builder
The Lesson Builder is the editor's planning and generation surface. It creates or updates a Lesson Plan, then writes the ordered Steps from the shared lesson brief, target language, and CEFR level.
Lesson and Canvas views
The authoring view switch is located on the floating bottom dock:
- Canvas (labeled Canvas or
Canvas · {count}) shows the structure and plan view with the Lesson Plan rail and Step cards. Use it to architect the lesson's flow, arrange steps, and manage timing. - Lesson shows the content editing view without the planning surfaces. Use it to read, write, and edit the lesson material students will see.
The mode switch tooltips explain whether the plan panel and Step cards or the student-facing content view will move into place. The Lesson Plan rail also has its own Hide the structure control, which switches to Lesson view; choose Canvas again in the floating bottom dock to bring the plan back.
The editor keeps a planned-but-empty lesson in Canvas view until a Step has content, even if Lesson was selected on another lesson. If the lesson has no plan, Canvas view shows an offer to create one. This keeps the only surface that can finish and write the plan available.
Self-study authoring uses the same Lesson / Canvas switch and the same Step controls.
How whole-lesson generation works
- Plan the lesson -- review the Lesson Plan's theme, situation, method, language, time, Stage labels, and ordered Steps using the Step Palette (opened by clicking the
+seam button between steps, at the bottom of the plan, or by clicking a Step header's kind chip). - Preflight cost estimates -- the sticky Live Rail calculates a real-time AI credit estimate broken down per Step, factoring in provider prompt caching discounts.
- Generate with zero-cost cancellation -- start whole-lesson or single-Step generation. You can abort mid-stream with zero cancellation penalty fees.
- Edit conflict fences -- if you type manual edits during generation, an edit conflict fence protects your document without overwriting your changes.
Steps are independent for completion. Some can finish while another is still writing or fails. The completed Steps remain available for review, and a failed Step can be written again without rewriting the rest.
Available Step types
The Step picker includes content and exercise types such as warm-up, vocabulary (with free cached pronunciation audio), theory, reading, Sentence Gap, listening, speaking, writing, games, brain breaks, and visual activities. Each Step supports up to 60 exercises configured directly in its card form.
Focused settings can include teacher-supplied pronunciation targets, handwriting upload limits, handwriting mistake categories, and Category Sort items. For the full workflow, see AI Lesson Generation.
Self-study lessons can include the Live Examiner speaking task. “Live” means a short audio role-play with the AI examiner, not a teacher joining the lesson. An ordinary speaking Step in self-study defaults to a recorded task, as described in AI Lesson Generation → Speaking Steps.
Review generated widgets
After AI generation, use Widget Edit Mode (the eye icon) on the floating bottom dock to switch all exercise widgets in the lesson into edit mode. This lets you review answers, options, instructions, scoring settings, and generated widget content quickly.
Click the eye icon again when you are done. Leaving widget edit mode on means the document is still showing teacher-edit controls instead of the normal lesson experience. You can also use Student Preview (the graduation cap icon on the floating bottom dock) to test how exercises behave from the student's perspective.
Some generated widgets, such as image prompts and audio dialogues, may start in edit mode on purpose. Review the prompt or script, make corrections, then click Generate inside the widget. If you do not generate the asset, the widget stays as an editable placeholder.
For widgets that depend on AI audio, such as Pronunciation Assessment, prepare and cache the generated audio before assigning the lesson when students should not trigger the AI cost themselves.
Before assigning the lesson, use Clear answers (the eraser icon on the floating bottom dock) to remove any test answers you entered during review. If test answers remain in the lesson document, students can receive those answers in their assigned copy.
Recover and retry generation
An empty or failed Step offers Write it again. To regenerate a Step that already contains content, open its actions menu and choose Replace this step. Confirm the warning before the existing content is removed.
If the connection drops, keep the editor tab open while it reconnects. Completed Step content that was already saved remains available. Check the Step cards after reconnecting and retry only the Step that is still empty or failed. If the editor shows a busy or waiting state, allow the current write to finish or use the available retry action later.
When a published lesson is generated, the result is reviewed in an unpublished Working Copy. Review the completed and retried Steps there, then publish the Working Copy when it is ready for students.
Always review AI-generated content before publishing. The AI can make factual, linguistic, or configuration mistakes, even when the workflow produces a coherent draft. The final material check remains the teacher's responsibility.
Ask AI Assistant
The Ask AI tool is a built-in assistant that helps with writing and editing tasks directly in the editor.
How to Use It
- Click the Ask AI button in the toolbar, or type
/askai - A prompt input appears where you can type your request. You can also paste, drag-and-drop, or click the attachment icon to include an image.
- The AI generates content, uses any attached image as context, and inserts the result at the cursor position (or directly onto the active slide if in Presentation Mode).
Working with Attached Images
Attaching an image means "look at this". The AI uses the picture as a reference for whatever you asked — describing it, building an exercise around it, writing a caption.
An attachment is never quietly turned into an edit. Your file is left alone, and asking in the prompt for the picture to be changed does not modify it. When you want a different picture, ask Ask AI to generate a new illustration instead; it returns an image you then insert into the lesson yourself. Generating an image uses AI credit when it is not covered by a plan allowance.
Retrying an Interrupted Answer
Every answer in the conversation shows how it ended — completed, stopped, or failed. A stopped or failed answer offers a retry.
Retry replays your original request exactly: the text you wrote, the materials and images you attached and the way you attached them, and your context settings. You never have to retype the request or pick the attachments again.
If an answer cannot be saved, you are told so and offered the retry.
Revising one part
When you ask Ask AI to revise one part of a lesson, it revises that part and leaves the rest of the document alone. The number of items you asked for, the columns, and the structure you specified are preserved.
Example Prompts
- "Write 5 example sentences using the present perfect tense"
- "Create a dialogue between two people ordering food at a restaurant"
- "Explain the difference between 'since' and 'for' at B1 level"
- "Generate a vocabulary list of 10 travel-related words with definitions"
- "Write a short story using these vocabulary words: adventure, mysterious, discover"
Ask AI generates content that fits your current topic and level.
Help Assistant vs. Ask AI
Locuto.PRO also includes a Help Assistant for documentation questions. Use it when you want to know how the platform works. Use Ask AI when you want to create or edit lesson content inside the Lex Editor.
Conversation Memory
Ask AI remembers context across messages. Once you specify topic, exercise type, level, or language, you don't have to repeat them.
For example:
- "Generate 10 vocabulary exercises on food for A2 level"
- "Now add images to each one" — knows you mean the food exercises at A2
- "Make 5 more but for B1" — carries over topic and type, updates the level
Fully specified requests generate content immediately without clarifying questions.
Lesson-Aware Requests
When you use Ask AI inside an existing lesson, the assistant automatically sees the lesson's name, description, CEFR level, lesson type, and tags. Short requests like "add a follow-up exercise" or "write a warm-up for this" work without you restating the topic. If you launch Ask AI from a live session, the student's age, gender, and native language are also available so the generated content can match the learner instead of a generic profile.
Ask AI with Materials Library
Ask AI can use your school's Materials Library as trusted source context. Upload or paste the source first, wait until the material is Ready, then use Ask AI from the Lex Editor to adapt that material into lesson content.
You can explicitly link specific documents to your request by typing @ directly in the Ask AI input. This opens a search menu where you can find and select documents from the Materials Library. Selected documents appear as chips in your prompt and strictly scope the AI's context to those specific files.
Good requests include:
- "Create a B1 reading lesson from this source with vocabulary and comprehension questions."
- "Summarize the material for A2 students and add a speaking task."
- "Turn this PDF passage into homework with deferred answer reveal."
- "Extract useful phrases from the source and make matching exercises."
- "Build a teacher-led warm-up from this image prompt."
When Ask AI uses library material, review the source references before publishing. This keeps the final lesson aligned with the uploaded material instead of drifting into a generic AI answer.
Citations stay out of what you insert. The content you put into the lesson is free of source markers. A short note listing the sources the answer drew on appears after it, in the chat only — so you can check where something came from without pasting reference numbers into the lesson your students read.
Ask AI with PDF Materials & Page Recognition
When you want to turn coursebook pages, worksheets, or exam handouts into interactive lesson steps, Ask AI can recognize and extract text, tables, and images directly from your school's PDF documents.
Opening the PDF Page Picker
You can open the PDF page chooser in three ways:
- Attachment menu: Click the attachment icon in the Ask AI input bar and choose Choose PDF pages.
- Type
@: Type@followed by the document title to search your school's Materials Library. Selecting a PDF opens the page selector. - Suggested chips: If your prompt mentions a school PDF by name, Ask AI offers a suggested library chip directly below the prompt input. Clicking the chip launches the page picker.
Two-Pane Page Chooser & Direct PDF Upload
The picker provides a two-pane layout designed for quick page selection and file upload:
- Left pane (Document list & upload): Lists available school library PDFs with a search bar (Find a PDF by name). Click "Upload PDF" right beside the search bar (or drag and drop a PDF file into the picker) to upload a new PDF from your computer. It is saved automatically to the school's Materials Library and opens immediately in the page picker.
- File limits and password support: PDFs up to 25 MB are supported (images and audio remain up to 15 MB). PDFs that have an owner password (files with print or copy restrictions that still open without typing a password) are accepted and unlocked automatically for page viewing and AI page recognition. Only PDFs that require a user password just to open are rejected ("This PDF needs a password to open. Upload a PDF that opens without a password.").
- Right pane (Preview & selection): Displays a scrollable thumbnail rail alongside a high-resolution page preview. You can:
- Browse pages using the thumbnail list or keyboard arrow keys (
LeftandRight). - Click Select this page to toggle individual pages.
- Type a page range (such as
1-3, 5or12-15) into the range box and click Apply. - Review the running count of selected pages and remaining allowance.
- Browse pages using the thumbnail list or keyboard arrow keys (
20-Page Limit per Conversation
Teaching staff can attach up to 20 pages total across all PDFs in a single Ask AI conversation. This limit keeps AI responses fast and focused on relevant lesson excerpts:
- If a document contains more pages than your remaining allowance, the Select all button is disabled with an explanatory tooltip prompting you to enter a specific page range instead.
- If you need material from another chapter later in your lesson design, you can remove existing chips or start a new conversation turn once earlier sections are drafted.
Per-Page Recognition Pricing & Free School Reuse
Recognizing PDF pages uses school AI Credits efficiently:
- New page recognition: Recognizing a new page costs 5 credits from the school's AI Credit balance (and counts toward your individual staff budget if one is set). The Ask AI response that follows is billed separately based on standard generation usage.
- Saved pages are free: Once any teacher or administrator in your school recognizes a page, that page is cached permanently across the organization. It displays a green Saved badge ("This page is already recognized. Using it is free."). Reusing saved pages costs 0 recognition credits.
- Transparent preflight estimate: Before you send your message, a preflight cost line appears below the composer showing exactly how many newly recognized pages will be billed and how many saved pages are included at no charge.
Attached Page Chips & Persistence
Selected pages appear as compact chips directly above the message composer (for example, Workbook · Pages 1–3, 5):
- Adjusting selection: Click an attached chip at any time to reopen the two-pane picker and add or remove pages.
- Turn persistence: Attached PDF chips remain active for follow-up turns in the conversation. You can ask follow-up questions (such as "Now convert exercise 3 on page 2 into a Sentence Gap widget") without re-attaching or re-recognizing the material.
- Removing context: Click the
×on a chip when you want to clear that PDF from subsequent prompts.
Background Processing & Partial Receipts
Page recognition runs in the background so you never lose work:
- Background resilience: Recognition continues even if you close the Ask AI drawer or navigate between lesson views.
- Partial receipts on interruption: If a network interruption occurs while processing multiple pages, any pages that finished recognizing are immediately saved and credited to your school library. A detailed partial receipt shows which pages succeeded.
- Continue request: An inline Continue request button lets you resume processing with one click, retrying only the remaining unrecognized pages without re-charging for the pages already saved.
Exercise Generation from Context
The AI can generate exercises automatically from your lesson content:
- Write or generate a text passage, grammar explanation, or vocabulary list
- In the exercise block configuration, choose "Generate exercises from context"
- The AI analyzes your content and creates relevant exercises:
- Fill-in-the-blank questions based on grammar patterns in the text
- Multiple choice questions based on vocabulary
- Matching exercises from vocabulary tables
- True/false statements about the reading passage
- Comprehension questions
Works well after a reading passage or grammar explanation.
Translation Panel
The translation panel lets you quickly translate text between languages from inside a lesson.
How to Use It
- Select the text you want to translate
- Click the Translate button in the floating toolbar (or right-click menu) to open the Translation panel
- Pick a language pair with the chips at the top of the panel
- Press Enter (or click translate) — the translation appears and you can insert it, copy it, or dismiss it
Language Pair Selector
The Translation panel offers quick chips for the language pairs most relevant to the current lesson:
- Student native ↔ course language — for a student-facing translation. In a course-lesson view that is not yet assigned to a specific student, this chip uses the course's default student-native language.
- Teacher native ↔ course language — for a teacher-facing reference.
Switching the pair does not auto-translate — you confirm with Enter (or the translate button) so the panel never spends AI credit on an accidental pair switch. If a translation is already in flight when you change the pair, the in-flight request is cancelled so the panel does not show stale output.
Use Cases
- Quickly translate vocabulary words for bilingual lessons
- Create side-by-side translation comparisons
- Verify your own translations
- Generate translations for matching exercises
Lesson Analysis
Beyond content creation, AI is also used to analyze student performance on completed lessons. This feature is accessed from the enrollment page rather than the editor.
What It Does
The AI examines a student's answers across all exercises in a completed lesson — fill-in-the-blank, multiple choice, matching, ordering, true/false, sentence builder, speech exercises, and essays. It produces a structured, evidence-based report with:
- Performance summary and exercise-by-exercise breakdown
- Error pattern identification — recurring mistakes across exercise types
- Strengths and weaknesses assessment
- Essay feedback with quality rating (if applicable)
- Follow-up recommendations — suggested topics, focus skills, and exercises
Reports are written in the student's target language.
How to Use It
- Go to the student's enrollment page
- Find a completed lesson (status: Passed or Verified)
- Click the brain icon — a confirmation dialog shows the cost
- The analysis runs in the background and a notification arrives when it finishes
- Click the yellow brain icon to view the report in a side panel
A bulk analysis option analyzes the 20 most recent lessons in a single request.
See View AI Analysis Reports for more details.
AI Costs
Some AI features use plan allowance first; otherwise they spend AI credit. The platform shows the billing source or required balance before paid generation.
For AI credit purchasing and balance management, see Billing & AI Credit.
Frequently Asked Questions
Can I use AI to generate homework from a class lesson?
Yes. Use Step-by-Step generation to create exercise sets based on topics you covered in class. You can also generate a full homework lesson using the same topic and vocabulary.
Can AI create audio content?
Yes. Listening Steps can be generated with text-to-speech. The AI writes a dialogue or monologue script, and the text-to-speech engine converts it to audio with configurable voices, emotional tags, and natural pacing.
Can the AI generate images?
Yes. Image Steps can include AI-generated images (realistic scenes or memes). The images are created during the content creation step and stored with your lesson.

