Changelog

✨ New Features

Bulk Actions for Courses

Admins can now select several courses at once in the course list and delete or update them in one go, with a confirmation step before anything changes. This makes it much quicker to clean up after switching from a test LMS to your live one, or after cross-course knowledge was turned on by mistake. You can update up to 100 courses at a time.

Google Drive as an External Tool

Google Docs and other Drive files that faculty link from their courses can now become course knowledge, in any LMS. Before, only Drive videos were picked up. To set it up, add Google Drive under Knowledge → LMS → External tools and share the relevant files or folders with the LearnWise service account. Content is only used in the course that links to it. Files LearnWise can’t access are shown to admins, and answers cite the original Drive link. Existing Google Drive video setups are moved over automatically.

🛠 Improvements

Mini-Game While Study Mode Questions Load

Learners can now play a short, relaxed mini-game while Study Mode prepares their questions. It has no score, ends on its own as soon as the questions are ready, and never takes time away from studying. Turn it on or off for each assistant in the Study Mode settings on the AI Tutor tab, under Loading experience. It’s on by default for new assistants and off for existing ones.

Higher-Quality Voices for Voice Mode, Now Including Italian

Voice Mode now offers a new set of natural-sounding HD voices. Under General settings → Voice mode, admins can listen to each voice and choose one for the assistant’s language. Speech is now processed in your organization’s hosting region, and Italian is now available as a Voice Mode language. Existing organizations are moved to the new voices by the LearnWise team, so contact your account manager if you’d like to switch.

File Attachments Available for Every Assistant

The File attachments setting under General settings is now available to all organizations, so each assistant decides for itself whether learners can upload files in the chat. The setting is off by default, so turn it on for each assistant where you want learners to be able to upload files. Files that learners upload can also be used in Study Mode.

Insights and Improvement Suggestions in Your Assistant’s Language

Trends topic titles and summaries in Insights are now written in your assistant’s language instead of English. The same applies to the titles and descriptions of AI-generated improvement suggestions.

Smoother Flow Building

New Flows are now created switched off, so you can build and test them before learners see them, then turn them on when they’re ready. Timezone dropdowns, including the one in the Flow schedule condition, now have a search box so you can find a timezone quickly.

EU AI Act Readiness

For organizations hosted in our EU region, AI-generated chat answers, workflow messages and study activities are now produced by models that add the machine-readable marking of AI-generated text that the EU AI Act requires from 2 December 2026.

✨ New Features

AI Help for Website Import Rules

Admins no longer need to write URL patterns by hand when adding a website as knowledge. Describe in your own words which pages you want (for example “only the admissions section”), and LearnWise writes the include and exclude rules for you. Beside the rules you see two lists, pages that would be imported and pages that would be skipped, so you only need to check that the lists look right. You can refine the rules with a further instruction, mark individual pages as in or out, or edit the rules by hand at any time. The card always says whether its example pages come from your last import, the site’s sitemap, or are AI-generated examples.

Note: Generating a new set of rules with this helper overwrites any existing include and exclude rules for that website.

This feature is disabled by default — reach out to your CX representative if you'd like it enabled for your organization.

Approvals for AI Ops Assistant Actions

Admins can now set a higher approval level for any AI Ops action, under Knowledge → Courses → your LMS integration → LMS Capabilities. When an action is set to high risk, users see a new high-risk approval card in the chat. It explains why the action is risky and asks them to review the details before they can allow it. Admins can also require approval for actions that only read data. Levels can be raised but never set lower than the default, so existing setups don’t change.

Data Retention Periods

Organizations can now decide how long conversation data is kept. A new Data retention tab in the organization settings, next to Audit log and Roles, lets you set a retention period of 30 days or more, for all assistants or only selected ones. When conversations pass that period, they are removed together with their messages, linked improvements and uploaded files, and each removal is recorded.

📄 How do I set data retention periods for my organization?

Data Processing Consent Screen

Institutions that need active student consent before the assistant processes their data can now turn on a consent screen for an assistant, under Assistant → General, under the Configuration section. When it’s on, students must accept before they can start chatting, and every acceptance is recorded. You can write your own consent message (including a link to your privacy policy) or use the default text. Signed-in students are asked again after one year, or whenever you change the consent text, and they can withdraw consent from the chat menu. The screen works in the standalone chat, the embedded widget, LMS launches and Microsoft Teams.

📄 How do I require learners to accept a data processing consent screen?

Anonymization per Assistant, with Identity Requests

Conversation anonymization can now be turned on for individual assistants, not just for the whole organization. When an assistant is anonymized, its conversations are anonymized for everyone in the Inbox, exports, Insights and the Teacher Dashboard, while other assistants stay identifiable. For cases like safeguarding or a learner complaint, a user can request the identity behind specific conversations, giving a reason. An admin approves or rejects the request, approved access is limited to the requester and those conversations, and every step is recorded in the audit log.

Nextcloud as a Knowledge Source

Nextcloud can now be connected as a knowledge source. Files stay in Nextcloud as the original, and when a file changes there, LearnWise picks up the latest version automatically, so admins don’t need to re-upload documents after every edit.

📄 How do I connect Nextcloud as a knowledge source?

🛠 Improvements

Danish Interface Language

Danish is now available as an interface language in the chat, the admin panel and AI Feedback.

Help Desk Escalation Updates

Help desks can now have planned closure periods, such as a holiday break, with a start and end date and an optional label. During a closure, escalation options are automatically shown as unavailable, so there’s no need to switch them off and back on by hand. For ServiceNow, different escalation flows on the same help desk can now create different ticket types (for example an incident for one flow and a case for another), each with its own field mapping.

More Course Content Formats

Assistants can now read course content from more e-learning formats, including Articulate Storyline and Adobe Captivate SCORM packages and more Articulate Rise packages. For Brightspace, syllabi published through the Syllabus Tool can now be used as course knowledge.

AI Feedback Updates

Students can now submit Apple Pages documents for AI Feedback. Admins can also give the feedback launcher button its own icon, either by uploading one or by reusing the chat launcher icon.

Longer Course Answering Style Instructions

Teachers and admins can now save much longer course-level answering style instructions, so there is room for detailed guidance such as a list of assessment questions the assistant shouldn’t answer.

Conversation Logs and Admin Tables

When a conversation is handed over from one assistant to another, it can now be found from both assistants, so teams that manage the receiving assistant can review the conversations it handled. Tables across the admin panel also now offer more consistent sorting, filtering and column selection.

Accessibility

Screen readers now announce controls in help desks, Improvements and the organization switcher with names that match what is shown on screen, making these areas easier to use with assistive technology.

✨ New Features

Custom Roles, Permissions, and Data Privacy Controls

Admins now have much finer control over who can see and do what in LearnWise. You can create custom roles (using an existing role as a starting point), choose exactly which areas each role can view or edit, and assign roles across the whole organization or to individual assistants. Every default role except Super Admin can now be assigned at the assistant level, and all role changes are recorded in an audit log.

You can also control who sees student identities. Roles can be set to see Inbox data in anonymized form only (with exports and Improvement previews anonymized too), and anonymization can now be switched on for individual assistants, which is useful when one assistant handles sensitive topics while others should keep showing full user details.

Managing people is simpler as well: the Members page shows which assistants each person is assigned to, pending invites can be cancelled, and the sign-in and account screens have a refreshed design.

📄 How do I create a custom role?

📄 How do I view the audit log?

Parent and Child Organizations

Institutions with several campuses, schools, or departments can now be set up as a parent organization with child organizations underneath. Parent organization admins get their own admin panel and dashboard, can view Insights across all child organizations, manage members of child organizations, and choose which features each child organization has access to.

📄 How do I manage child organizations from a parent organization?

Smarter Course Content

Course assistants can now follow links inside course pages to files (PDFs, Word documents, PowerPoints) and other pages within the same course, and add that content to their knowledge. Previously, linked material was skipped, which could lead to “I don’t know” answers. PDFs added to Moodle courses as URL links are now read too. Admins can also filter knowledge sources by whether their automatic updates are paused or active.

Microsoft Dynamics 365 as a User Data Source

Microsoft Dynamics 365 can now be connected as an external user data source, alongside existing options like Microsoft Entra. Admins choose which profile fields to pull in, and those details become available in conversation context and Flows, so the assistant can tailor its answers to each user.

🛠 Improvements

Help Desk Escalation Updates

Help desk hours can now include more than one time window per day (for example, 9:00 to 12:00 and 14:00 to 17:00), so escalation options only appear when staff are actually available. When a learner opens a help desk from a conversation, the Back button now takes them back to that conversation. Escalation tickets sent to TeamDynamix can also include file attachments.

Hide Shortcut Buttons

Shortcut buttons can now be hidden without deleting them, so you can take a button out of the chat temporarily and bring it back later with the same settings.

Chemistry, Music, and Math Notation in Chat

The chat can now display chemical structures and musical symbols, and more math expressions now display correctly, making assistants more useful for science, math, and music courses.

Faster Browsing in Grading Studio

The courses list and each course’s assignments list in Grading Studio are now split into pages, so large catalogs load faster.

Update Microsoft Entra Credentials

Admins can now update the credentials of an existing Microsoft Entra connection instead of having to set it up again.

Accessibility

Voice Mode now confirms out loud what it heard before sending, and reads the latest response first, making it easier to use for learners who rely on audio.

✨ New Features

Grading Studio

Grading Studio is a dedicated space for reviewing and grading student submissions: an inbox of submissions organized by status, a built-in document viewer, and three ways to start feedback (type notes for AI to expand into full comments, record a voice note, or generate feedback straight from the rubric). Instructors can edit the AI-generated feedback freely, rate each rubric section as helpful or not, and add inline annotations directly on the submission.

Grading Studio is currently available for Microsoft Teams, with more LMSs supported in the future.

📄 What is the Grading Studio and how do I set it up?

Exclude Conditions in Flows

Admins building Flows can now set a condition to exclude specific values, not just match them, across role, course, external data, and URL conditions.

🛠 Improvements

More Flexible Shortcut Buttons

Admins can now assign specific roles to individual shortcut buttons, so different groups (students, staff, teachers) each see the shortcuts most relevant to them. Admins can also enable shortcut buttons inside an active conversation, not just the welcome screen.

HEIC/HEIF Image Support in Feedback

The feedback tool can now read HEIC and HEIF image files (the default photo format on iPhones) in student submissions.

Anonymized Conversation Logs

Admins and instructors can now turn on anonymized conversation logs in the Admin Inbox and Teacher Insights, showing "Anonymous User" instead of real names and scrubbing personal details, including a student's approximate city and country. This also applies to conversation messages themselves.

Copy Button for AI Answers

Chat now includes a copy button under each AI answer that preserves formatting, so tables and text can be pasted directly into a document or spreadsheet.

Control Automatic Updates for Website Knowledge Sources

Admins can now choose whether a website or web page knowledge source updates automatically, both at creation and afterward from the knowledge table.

Real Video Titles in Source Citations

Video sources cited in chat answers now show the actual video title instead of a generic ID.

Respect Brightspace Access Rules for Past Courses

If your Brightspace settings restrict students from accessing past or closed courses, Tutor's cross-course chat now follows those same rules.

Tutor Insights Section Isolation (Brightspace)

For Brightspace institutions running many sections under one course, Tutor Insights now shows each instructor only the conversations from their own section.

🩹 Patches

The "Adjust" Button Now Follows Instructions More Closely

When instructors ask the Feedback tool to adjust a piece of feedback, it now follows their instruction much more closely. If you ask it to change one specific thing, that's the change it makes.

Teacher Instructions Take Priority

When teacher-level instructions conflict with admin-level instructions, the teacher's instructions now win. Previously admin instructions took priority, but since the instructor is the one giving the feedback, they should be in control of it. This also means requests like "make it shorter" are now reliably respected.

One Set of Instructions per Automated Feedback Run

Automated assessment now uses only the instructions set in its own custom instructions. Previously, instructions from other places could carry over into automated runs, which sometimes led to feedback that didn't match what was configured.

Regenerating Feedback Keeps the Same Grades

Regenerating feedback now builds on what was generated before, so results stay consistent instead of starting from scratch, unless you ask for something different when regenerating. Grades also stay the same unless the submission or rubric changes. To change a grade after generation, instructors can edit it manually.

Longer Custom Instructions

Custom instructions can now be up to 5,000 characters, giving admins and instructors more room to describe how feedback should be written.

🛠 Improvements

Instructor Grading Time Insights

Admins can now see how much active time instructors spend grading and giving feedback, right in Insights — at the assignment, session, and individual-submission level. Time is tracked only while an instructor is actively working in the feedback tool (idle periods and tab-switching don't count), making it easier to show the time AI Feedback is actually saving.

Cross-Course Chat: Canvas Sub-Account Filtering

Cross-course chat can now be scoped to one or more Canvas sub-accounts, with the option to include nested sub-accounts — not just the whole instance. This can be combined with CSV-based filtering.

Freshdesk Escalation: Custom Field Mapping

The Freshdesk escalation integration now reads your instance's actual custom ticket fields and their real dropdown values (ticket type, status, priority, etc.), so admins can map required fields correctly instead of choosing from generic defaults. This also fixes tickets being rejected outright when an institution has customized its Freshdesk field values.

Cleaner Conversation Exports

Admins can opt in to a cleaned-up export format for conversations that strips out internal "thinking" steps and matches what's shown in the Inbox — making it much easier to run outside reporting or analysis on exported conversation data.

Expanded Improvements Export

The Improvements export (CSV and JSON) now includes Assistant ID, Conversation ID, Student Name, Student Email, and Student ID columns, making it easier to tie an exported item back to a specific assistant, conversation, or student.

Richer Escalation Metadata

Escalations sent to Slack, Teams, Google Chat, or a connected ticketing system now include the conversation's end time and per-message timestamps, not just the start time — useful for understanding how long a learner spent in a conversation before escalating.