Product roadmap

Choose what information an Assistant uses to answer, whether that's the web, your whole Knowledge Center, or just the assets you pick.

Specify where each project gets its information, such as from the web, the Knowledge Center, or specific assets.

Upload one or more files from your device and use them in your conversation with the Assistant.

Set a command to run automatically on a schedule, like generating a weekly report or sending monthly reminders to anyone with overdue training.

Generate natural-sounding audio, like a voiceover, just by describing what you need or sharing a script.

Create short videos from a simple description, such as a welcome message for new hires or a quick explainer.

Save a canvas from your chat with the Assistant as a FlexDoc in the Knowledge Center, where you can share, edit, and assign it like any other asset.

Use the Assistant inside other apps, such as SmarterU LMS, so you can ask questions without leaving the tool you're in.

Paste longer or more varied content into the chat and have it handled cleanly, so your message stays easy to read and the Assistant gets everything you shared.

Hide the tips shown on pinned commands once you know how to use them.

Work on longer content, such as a document draft, in a canvas beside the chat. Ask the Assistant to make changes as you go.

Ingest a document you already have — a policy PDF, a slide deck, a product guide, or a recorded call — and automatically structure it into a course outline with bite-sized lessons, a knowledge check, and activities. Generation is grounded only in approved sources, and every draft is fully editable before it ships.

A family of live content blocks that present material inside a lesson — rich text, charts, KPI metrics, dashboards, Mermaid and interactive flow diagrams, tabs, tables, comparison tables, images, video, code, flashcards, forms, callouts, and sandboxed HTML — all generated from the author's own documents.

Seventeen self-scoring question types — multi-question quiz engine, single- and multi-select, true/false, fill-in-the-blank, short answer, matching, categorization/sorting, hot spots, labelling, and ordering — all generated from source material with instant, colour-coded feedback.

Six AI-driven, source-grounded practice formats: Knowledge Coach (Socratic walkthrough), Conversational Scenario (in-character dialogue), Situational Scenario (step-by-step problem solving), Data Entry Scenario (field-by-field form practice), Software Simulation (watch-try-prove-it from screenshots), and a Custom guided experience. All can be graded or ungraded and adapt to what the learner does.

When no template fits, a coded custom interactivity — a deterministic, browser-based widget or simulation — can be scoped and built for a specific course.

Five slash commands a learner types to start a session: /learn to work a concept out, /quiz to check what stuck, /memorize to drill the details, /scenario to make the calls, and /roleplay to practice the conversation. Each runs in the chat thread, grounded in the organization's own documents.

Every course ships with a Presentation mode that draws content onto the screen in sync with AI-generated narration in a chosen voice and language — describing images, summarizing tables, and revealing bullets as it speaks. Progress is tracked at the element level, so a half-finished course resumes at the exact point.

Each course includes a page-aware Learning Assistant docked alongside the content that can summarize the current page, quiz the learner on it, surface takeaways, or explain concepts. It travels with the SCORM export into any LMS.

When a learner fails a quiz they are offered a one-on-one coaching session on the exact pages they missed before retrying, rather than hitting a dead end. Quizzes support three depths — recognition, open-ended AI-graded recall, and a recall challenge.

Generate the same course — quizzes, roleplays, and coaching included — in dozens of languages from one source document, without a manual translation pass. The source can be one language while quizzes and answers are produced in others.

Apply themes and organizational branding to a course so its look matches the company, configured once and applied across the course.

Export a finished course as a SCORM or xAPI package in one click and import it into any standards-compliant LMS with no re-authoring or migration. SCORM is the broadly compatible default for completion and scores; xAPI captures richer activity detail.

SOC 2 Type II attestation covering security, availability, and confidentiality controls. On top of the encryption, grounding, and audit controls already in place.

ISO 27001 certification of the information-security management system. On top of the encryption, grounding, and audit controls already in place.

GDPR compliance for handling and protecting the personal data of EU data subjects. On top of the encryption, grounding, and audit controls already in place.

HIPAA compliance for safeguarding protected health information (PHI). On top of the encryption, grounding, and audit controls already in place.

A self-scoring interactivity generated from your source material. Runs a multi-question quiz with a progress bar, optional timer, per-question scoring, and a results screen. Learners get instant, colour-coded feedback and the result feeds completion and scoring.

A self-scoring interactivity generated from your source material. Presents options where exactly one answer is correct. Learners get instant, colour-coded feedback and the result feeds completion and scoring.

A self-scoring interactivity generated from your source material. Presents options where several answers may be correct, scored all-or-none or with partial credit. Learners get instant, colour-coded feedback and the result feeds completion and scoring.

A self-scoring interactivity generated from your source material. Presents a statement for the learner to judge true or false. Learners get instant, colour-coded feedback and the result feeds completion and scoring.

A self-scoring interactivity generated from your source material. Asks the learner to type the missing word or phrase into a blank. Learners get instant, colour-coded feedback and the result feeds completion and scoring.

A self-scoring interactivity generated from your source material. Accepts a free-text response and grades it with multiple scoring styles. Learners get instant, colour-coded feedback and the result feeds completion and scoring.

A self-scoring interactivity generated from your source material. Asks the learner to pair items from two sets. Learners get instant, colour-coded feedback and the result feeds completion and scoring.

A self-scoring interactivity generated from your source material. Asks the learner to sort items into the correct categories, via dropdown or drag-and-drop. Learners get instant, colour-coded feedback and the result feeds completion and scoring.

A self-scoring interactivity generated from your source material. Asks the learner to select the correct region of an image or span of text. Learners get instant, colour-coded feedback and the result feeds completion and scoring.

A self-scoring interactivity generated from your source material. Asks the learner to label parts of an image or diagram. Learners get instant, colour-coded feedback and the result feeds completion and scoring.

A self-scoring interactivity generated from your source material. Asks the learner to arrange items into the correct order, via dropdown or drag-and-drop. Learners get instant, colour-coded feedback and the result feeds completion and scoring.

An AI-driven, source-grounded practice format. Walks the learner section by section with Socratic questioning until they can explain the material back. It can be graded or ungraded and adapts to what the learner says or does.

An AI-driven, source-grounded practice format. Puts the learner in an in-character dialogue and gives coaching feedback after each reply. It can be graded or ungraded and adapts to what the learner says or does.

An AI-driven, source-grounded practice format. Steps the learner through a non-conversational problem, decision by decision. It can be graded or ungraded and adapts to what the learner says or does.

An AI-driven, source-grounded practice format. Has the learner complete a form field by field with live validation. It can be graded or ungraded and adapts to what the learner says or does.

An AI-driven, source-grounded practice format. Builds a watch-try-prove-it walkthrough of a workflow from uploaded screenshots. It can be graded or ungraded and adapts to what the learner says or does.

An AI-driven, source-grounded practice format. Runs a bespoke Assistant-style conversation configured with the author's own AI instructions. It can be graded or ungraded and adapts to what the learner says or does.

Content blocks that render bar, line, pie, doughnut, scatter, and radar charts inside a lesson, generated from the author's documents and data.

Content blocks that present KPI metrics and mini dashboards within a lesson, drawn from source material.

Content blocks that render Mermaid and interactive flow diagrams inside a lesson to explain processes and relationships.

Content blocks that lay out side-by-side comparison tables generated from the author's content.

Content blocks that render custom, sandboxed HTML/CSS widgets safely within a lesson.

Flag an Assistant's answer as incorrect or missing information, so it can be reviewed and improved.

Paste a file into the chat and ask about it. Get a quick summary of a report, or have a policy draft checked for gaps.

Choose from ready-made commands, such as Summarize or Generate Quiz, to perform common tasks in one step. Pick which sets of commands each Assistant offers, so everyone gets the ones most relevant to their work.

Pin a command so it stays active as you keep chatting, until you unpin it. You'll always see what's pinned at the top of the chat, and you can still run other commands alongside it.

Choose how much thinking a command gets, so quick tasks come back fast and bigger ones get deeper reasoning.

Each FlexDoc has a private draft and a published version. Admins create a draft from a published doc, edit with auto-save, switch between draft and published views, and make changes public with Publish — with only one draft at a time.

Every asset carries an admin panel (Overview and Details) exposing title, active/inactive status (inactive is admin-only), an editable About description, created/edited metadata, and an API ID for programmatic reference.

Learners use /roleplay to rehearse a tough conversation against a realistic AI counterpart that stays in character — for example handling a price objection — with coaching feedback after every reply, all grounded in the company's own approved documents.

The /learn command starts a Socratic coaching session that asks guiding questions and offers hints rather than handing over the answer, continuing turn by turn until the learner can explain the concept back — with every turn cited to the learner's own documents.

The /quiz command generates a multi-question quiz on a chosen topic pulled straight from the learner's source material, with progress, scoring, and explanations that cite the source for each answer. No setup, startable anytime.

The /scenario command puts a learner into a situation that responds to the calls they make, tracking what they have ruled out as they work it to a resolution. What pushes back is a system rather than a person, so the decision is what gets assessed, in a debrief at the end.

The /memorize command drills a fixed set of terms, figures and steps until each one comes back from memory twice, spaced apart. A miss brings an item round again sooner, and nothing retires until it has been produced without help.

Send several assets at once to many people or whole teams, each with its own due date, warning period, and acceptance window. The acceptance window credits anyone who already completed the same asset within a set timeframe so they aren't assigned it again.

Define completion as a passing assessment score rather than a click, so a course is only 'done' when mastery is proven. Certificate templates are designed once and assigned per asset; on completion a branded certificate is issued with the learner's real data and appears in Tasks to download.

Completions can carry a valid-until date, so when validity lapses the task automatically returns for recertification rather than requiring someone to remember and re-assign it.

On completing an assignment, learners receive an AI-written debrief summarizing how they did and what to revisit, and Joy's Learning Assistants leave progress notes and an end-of-task summary on each assignment.

Live analytics show completion rates, average quiz scores, and at-risk learners at a glance, broken down by person and group, with exportable reports for audits or stakeholders. The same completion record is kept in Joy whether a course ran natively or via an exported SCORM package in the LMS.

Connect business tools (CRM, support, project systems), HRIS, corporate and academic LMS platforms, and collaboration tools so their content becomes searchable alongside the Knowledge Center from one place — and ask questions that span several systems at once for a single grounded, cited answer.

A catalog of 73 pre-built connectors spanning communication, CRM, data, e-sign, finance, HR, knowledge, LMS, marketing, project management, support, survey, and video tools — among them Salesforce, Slack, SharePoint, Google Drive, Workday, Jira, Zendesk, and HubSpot. Each one is added without custom development. Browse the full list in the connector library.

Connectors are read-only and permission-aware by default — each user only sees data they are already authorized to access in the source system, and no source system is modified.

A guided flow for connecting a system — authenticating, selecting what to make searchable, and bringing that content into unified search alongside the Knowledge Center.

A native connector for SmarterU LMS that goes beyond making content searchable to make the LMS conversational — ask about enrollments, completions, and learner activity in plain language and get grounded answers. Permission-aware and read-only, it works alongside the LMS rather than replacing it.

Connect your own services and systems beyond the pre-built library through REST APIs, GraphQL, webhooks, and other interfaces, extending cross-system search to in-house or niche tools under the same read-only, permission-aware model.

Optional write capability that lets a connector take an action in a source system (rather than only reading from it), enabled only where IT explicitly approves.

A more flexible permissions system beyond the current two roles, allowing finer-grained control over what individuals and groups can see and do.

Enterprise single sign-on with the customer's identity provider, with server-verified identity and on-demand provisioning when embedded in an LMS — so users reach JoySuite through existing credentials.

Data is encrypted in transit and at rest with 256-bit encryption, and customer content is never used to train AI models, with retention kept under customer control.

Answers are restricted to admin-approved sources rather than the open web, every answer is cited and traceable for audit, and audit logging covers administrative actions.

Ask a question in your own words and get an answer from content in the Knowledge Center, and from the web when it helps. Every answer shows its sources, so you can check where it came from.

See helpful tips while the Assistant works on your answer, such as what you can ask or which commands to try. Admins can add their own tips for each Assistant, or turn tips off.

Create specialized Assistants for different teams and topics, like an HR Assistant for policy questions or an Onboarding Coach for new hires. Give each one its own name, title, picture, and instructions, then assign it to the people who need it.

Continue any chat where you left off. Every chat is saved automatically and named after your first message. You can rename, favorite, archive, or delete chats to keep your history tidy.

Ask questions in the language you're most comfortable with and get answers in that same language, even when your content is written in English. You can also ask the Assistant to translate existing content into another language.

Copy the Assistant's response and paste it cleanly into an email or document. Links that only work inside Joy are removed automatically.

Save the requests you make most often as commands that you and your team can pick from a list, instead of typing them out each time. For example, create a Summarize Meeting command, then use it to turn meeting notes into a short summary.

Create downloadable Word, PowerPoint, Excel, and PDF documents by telling the Assistant what you need. For example, ask for a PowerPoint deck that introduces a new policy, or a PDF checklist for onboarding a new hire.

Describe the image you want and the Assistant creates it, from simple illustrations to photos and diagrams with readable text. Ask for changes until it looks right.

Get answers in the layout that fits best, such as tables, charts, timelines, flowcharts, mind maps, or forms. To get a specific format, just name it in your question.

Get charts, key numbers, and tables side by side. For example, ask for a dashboard of a sales spreadsheet in your Knowledge Center, then keep chatting to adjust what it shows.

The central hub where all knowledge assets — files and FlexDocs — are stored, organized into folders, and displayed as cards. It provides intelligent, meaning-based search beyond keywords and serves as the grounded information source that fuels the Knowledge Assistant.

Organize assets into folders nested up to 10 levels deep, shown as a left-hand tree and as cards. Folders inherit permissions from their parent, can be created, renamed, and moved by drag-and-drop, and are navigable via tree menu or breadcrumbs.

Upload assets — PDF, DOCX, XLSX, CSV, TXT, PNG, JPEG, MP3, MP4, WEBM — individually or in bulk. Files can be previewed, downloaded, assigned as tasks, managed through an admin panel (status, description, API ID), and queried directly through the Knowledge Assistant.

Customizable, collaborative documents made of pages containing text, images, links, embedded media, and tables — used for policy manuals, SOPs, and internal guides. They support full-screen viewing, embedded video, and assignment as tasks, and feed the Knowledge Assistant.

A block-based authoring tool with a slash menu for inserting content — headings, lists, tables, columns, accordions, images, embedded YouTube/Vimeo/Figma, LaTeX equations, Excalidraw whiteboards, and voice dictation. Built-in AI generates content from a prompt and refines highlighted text (improve, summarize, expand, fix grammar), each with accept/reject.

Organize a FlexDoc into pages like chapters, listed in a tree menu and nestable into a hierarchy up to 6 levels deep. Pages can be added, renamed, reordered by dragging, and nested or deleted.

Assign any Knowledge Center asset — docs, PDFs, videos, guided experiences, or full courses — to individuals, teams, or the whole company. Each assignment becomes a task on the recipient's Tasks page, with a subject and optional message; selecting multiple assets creates a separate task per asset.

Track tasks on a Kanban board across five stages — Not Started, In Progress, Warning, Overdue, Completed — with My Tasks, Assigned by Me, and All Tasks views. A configurable warning period flags tasks before their due date, and admins see every task in the org.

Centralized administration where admins add, view, edit, search, and inactivate users and manage detailed profiles (contact info, coordinators, timezone, preferred language, API ID). The account must always retain at least one Admin, and inactive users cannot log in.

Access is governed by two roles applied across the workspace: Admin (full administrative access) and User (a streamlined day-to-day experience).

Ask for a chart, such as a bar or pie chart, or let the Assistant choose one when it's the clearest way to show the answer.

See key numbers as easy-to-read cards. For example, ask for a summary of a report and get its main figures, such as total sales or a completion rate, each in its own card.

Get flowcharts, Gantt charts, mind maps, and other diagrams in your answers, so you can see how a process works or how the pieces fit together.

Get long answers broken into tables, tabs, and sections instead of one big block of text.

Get answers you can interact with, like flashcards to study from or side-by-side comparisons you can click through.

Built into the editor: generate content from a prompt (Write article, Create outline, Write blog post) and refine highlighted text (Improve writing, Make formal/casual, Summarize, Expand, Fix grammar, Explain concept), each with accept/reject.

Embed YouTube, Vimeo, and Twitch video, plus Tweet and Figma embeds, directly inside a FlexDoc, alongside images with alt text and captions.

Insert LaTeX equations and Excalidraw whiteboard diagrams as blocks within a FlexDoc.

Dictate content into a FlexDoc using built-in speech-to-text.