Institutions can now connect external data sources to enrich conversation context. Supported sources: CSV upload, Microsoft Entra, and LTI injector parameters. Fields (e.g. student program, campus, enrollment status) can be mapped, aliased, and surfaced in the Inbox. Flow conditions and custom message variables will come in a future release.

Math expressions now render using the MathML standard, enabling screen readers (VoiceOver, NVDA, JAWS) to read them properly instead of outputting raw markup. You may also notice some early improvements to how math is displayed visually in chat, with a more comprehensive update to follow in a future release.
You can now hand off a conversation from one assistant to another directly within a flow, allowing you to route users to a more specialized assistant based on their needs.

PDFs stored in the Brightspace Learning Object Repository and embedded in courses are now picked up during knowledge ingestion.
Flow-triggered emails now support a much richer set of template variables: conversation summary, history, user location, launch URL, IP address, and other system metadata.

Admins can now set a custom background color for the chat header. Text and buttons adapt automatically for readability. The Style tab layout has also been cleaned up with better labeling and spacing.

Voice mode now supports multilingual dictation and audio playback, removing the previous English-only constraint.
Several updates this release: automated feedback now has its own independent "Submit as draft" setting; automated entries show "Automated" as the teacher name in the dashboard and exports; teacher thumbs up/down ratings are now included in data exports; and Feedback V2 assistants show an animated thinking steps view during post-generation.
LearnWise now reasons through questions before responding. Rather than retrieving the closest-matching answer, the assistant breaks down each query, evaluates context, runs multiple search passes, and builds a response step by step.

What's changed:
Transparent reasoning β The assistant shows the search steps it took and the logic behind its answer, visible to users directly in the chat.
Better intent resolution β Vague or layered questions are worked through using conversation context (course, prior messages, likely intent) before a response is generated, reducing generic answers.
More concise responses β Average response length has dropped from ~1,990 to ~730 characters. More upfront reasoning produces tighter, more direct answers.
Graceful handling of knowledge gaps β Instead of "No sources found", the assistant explains what it found, what's missing, and asks for clarification on ambiguous queries.
Full guardrail compliance β Out-of-scope deflection, enrollment workflows, and academic integrity boundaries are fully upheld
Agentic Search is now the default for all institutions as of April 9th.
π βHow does Agentic Search work?β
The Flows section has been fully redesigned.

All flows now appear in a single flat list with inline drag-and-drop reordering, replacing the separate Prioritize modal.
New actions are available: Enter Study Mode, Detect and Start Interactive (with auto/manual toggle), and a Follow-up Questions action that supports both AI-generated and manual modes.
The General settings tab has been reorganised into a cleaner Configuration section, and Answering Guidelines has been renamed to Answering Style.
π βWhat are Flows and how can I create & manage them?β
CSAT ratings (1β5) and written user feedback are now surfaced directly in the Admin Inbox, visible on conversation rows in the list view and in full when opening a conversation.
π βWhat is the CSAT survey and where does it appear?β

Admins can now refresh or update credentials for OAuth-based integrations (e.g. HubSpot, SharePoint, Panopto) directly from the assistant knowledge settings, without needing to reconnect from scratch.

Admins can now set and update a timezone for their organization. All timestamps, date filters, and schedules across Insights, Inbox, Feedback, Exports, and Notifications now reflect the org's local timezone. New organizations can select their timezone during onboarding; existing organizations default to UTC.

The 300-character limit on positive and negative search filter fields for website knowledge sources has been increased, and the error message now clearly identifies which field is affected.
When an escalation form is connected to an integration with required mapped fields, those fields are now explicitly shown in the admin UI rather than surfacing a generic error message.
The feedback pipeline has been rebuilt from the ground up. The AI now analyses each part of a submission individually, producing more specific feedback and more accurate grading suggestions. Generation is faster on average, as the tool starts preparing in the background before the teacher opens it. Teachers set their personalization preferences once on first launch and they apply to all future sessions. Video and multi-file submissions are now supported.
π βWhat is AI Feedback & Grader?β
π βHow should instructors use AI Feedback & Grader?β
Note: The latest version of Feedback & Grader is rolling out gradually. Your Customer Success contact will confirm when it's available for your institution.

The LearnWise admin panel now has a new sidebar navigation and cleaner layout. Within each assistant, Study Mode and H5P settings have moved to a dedicated AI Tutor tab and AI Feedback now has its own tab, replacing the old Modes section. Cross-course settings now live under Knowledge, and the LearnWise help assistant is accessible from directly within the panel.

LMS credentials (Canvas, Moodle, etc.) can now be set up once at organization level and reused across multiple assistants, rather than configured separately for each one.
Students using LearnWise in MS Teams or Blackboard can now ask questions across all their enrolled courses in a single conversation. Course context is clearly shown when the assistant hands a conversation to a more specialized assistant.
Admins can now choose which specific team or queue a ticket goes to when setting up an escalation to Freshdesk, ServiceNow, or HubSpot.
Admins can now add or remove individual pages from an existing web knowledge source without deleting and re-configuring it. Fetch files sources now show which files have been found so admins can choose what to include.
French (France), German, and Japanese are now available as chat UI languages.
Admins can now be assigned permissions on a per-assistant basis, rather than only at the organization level. This lets institutions give specific staff access to manage individual assistants, for example different departments or teams managing their own assistants independently, without granting org-wide access.
π βWhat are the different roles that are available in the LearnWise dashboard?β

A new workflow action that automatically sends an email when a workflow is triggered, no API setup required. Admins configure the recipient, subject, and body, and can insert dynamic fields like the user's name, email, course name, and course ID. Useful for scenarios like student-at-risk alerts, where a counsellor or staff member needs to be notified when a student flags they're struggling.
π βWhat are triggers and their corresponding responses in Flows?β

Admins can now set a custom font size for the assistant chat interface. The setting is available in the admin panel under Assistants > Style, and it applies to chat messages, suggested questions, and follow-ups.
Admins can now manually reorder escalation methods and fields within help desks without having to delete and recreate them.
π βHow to connect my assistant to a help desk?β

The assistant can now connect to Microsoft Teams via API, giving students access to live course data including assignment deadlines, course content, announcements, and calendar events directly within their conversations.
Flow buttons now support two new action types. "Send Text Into Chat" inserts a prompt into the chat on click, and "FAQ" displays a saved FAQ answer directly in the conversation. This lets admins build flows that guide students with follow-up prompts and instant answers.

We have been working through a third-party accessibility audit to bring the end-user interface up to WCAG 2.2 AA compliance. This release includes a batch of fixes from that audit, improving the experience for students using assistive technologies.
Admins can now upload files (PDF, DOC, DOCX) directly to specific courses via a new Course Files tab in the assistant knowledge settings, without needing a full course scan. Files can be tagged to all courses or specific courses only.

Admins can now duplicate an existing assistant from the All Assistants screen. General settings, styles, flows, and FAQs are copied to the new assistant. Note that knowledge sources and help desks are not copied at this stage.

The improvements dashboard now supports filtering by LMS course ID. Filters are also applied when exporting, so the Course ID column will be reflected in exports.
Admins can now set a timezone when configuring Help Desk escalation availability. The end user will always see opening times converted to their own local timezone.

When escalating to a Help Desk, admins can now choose to push specific parts of the conversation into ticket fields, including a chat summary, full conversation history, and individual pieces of user metadata (e.g. name, email).

Reduced delay in response rendering so answers appear faster, particularly for longer responses.
Several improvements to the feedback generation pipeline, including sharper feedback quality with better rubric alignment and more reliable video submission handling.
The limit on shortcut buttons has been increased to 50. When more than five buttons are configured, they are grouped under a "More" option in the chat interface to keep things clean.

Several accessibility fixes across the end-user front-end, including support for 200% zoom with correct text magnification and improvements for screen reader users.
Media files embedded in Moodle online text submissions are now processed and included in feedback generation.
The Teaching and Learning Assistant can now be configured to ignore student-generated posts across all MS Teams channels and announcements.
Restrict specific knowledge sources to certain user roles in websites, applications, and files. When users authenticate, they only see content relevant to their role. For example, restrict instructor-facing ServiceNow articles to faculty while keeping student support articles visible to everyone.

Control which courses can access your assistant through LTI by uploading a CSV of allowed or blocked course IDs. This enables controlled pilot rollouts to specific courses without managing individual LTI placements. Particularly useful for Blackboard UEF integrations where placement control is limited.

Improved AI-generated feedback for video submissions to accurately evaluate both content quality and presentation skills, ensuring meaningful assessment of video assignments.
Added Dutch, French, German, Irish, and Welsh language options for chat interface translation.
Escalation options now show when support is available. If currently closed, users see the next available day and time, helping them know when to expect help.

The feedback tool now supports .tex files for institutions with math-heavy or technical coursework.
Added toggle to enable or disable content-level access checking in Moodle, giving you control over whether the assistant respects module restrictions and availability conditions.
You can now resize the chat window horizontally by dragging the edge, giving you control over the width to fit your workflow needs.

Configure API POST requests directly within Help Desks and Flows to integrate with external systems (e.g., SIS, CRM, and automation tools). Supports multiple authentication methods (API Key, Bearer token, JWT Bearer, Basic Auth) and dynamic variables like user information and conversation data.

Feedback generation now offers more granular length control with four options: Short, Medium, Long, and Unrestricted, giving you better control over feedback detail.

Implemented dynamic model selection as a switchable feature to optimize performance and cost based on task complexity.
When only one rubric exists, it now automatically expands in the grading interface, reducing clicks and improving grading efficiency.
Mandatory fields (name, description, email) have been removed when escalating to a phone number or external URL, creating a faster support experience.
iFrame workflow actions now support custom height settings, allowing better control over embedded content display.

The feedback tool now gracefully handles unsupported file types (e.g. ODT) with clear messaging.
Filter states are now consistently cached across Inbox, Improvements, and Feedback Insights pages.
Conversations now start fresh after 1 hour of inactivity or when starting a new browser session, ensuring cleaner context and more accurate responses for each topic.
Organizations can now hide user metadata (name, email, LMS User ID, IP address) in Inbox and Teacher Insights for enhanced privacy. Contact your LearnWise representative to enable this feature.

Added country and city filters to Inbox, making conversation data more actionable for reporting and analysis.

Improved query performance for Inbox conversations, resulting in faster load times for both the Inbox page and teacher dashboard.
CSV exports now open correctly in Microsoft Excel without formatting issues for complex conversations containing tables or multi-line content.