Requests

This feature enables secure user context sharing between your application and the assistant experience. Organizations will be able to pass authenticated user information from a logged-in environment directly into the injector using a cryptographically signed payload. This allows assistants to deliver more personalized and context-aware experiences while maintaining security and trust in the transmitted user data.

This update introduces centralized knowledge management across your organization. Shared knowledge sources — such as websites, applications, files, and FAQs — can be reused across multiple assistants, reducing duplication and making updates easier to manage. It also improves visibility into knowledge health by surfacing sync and update issues directly in the admin interface, helping teams quickly identify and resolve problems with sources like web scrapers or integrations.

Automatically deploy to an EC2 instance when a feature branch is marked as "QA ready" so that QA can be done on that feature

Remediate all WCAG 2.2 Level A and AA criteria marked "Partially Supports" in the V1.41 VPAT audit (April 2026).

Unify H5P interactive widgets with Study Mode so the tutor can mix text questions and H5P exercises in one adaptive quiz flow, and phase out Smart Triggering.

Enable LearnWise to integrate with major ticketing systems used by institutions

This update introduces improved support for external data in workflows. External data can now be used as a dedicated condition within workflow logic, allowing for more precise branching based on external inputs. In addition, external data can be used as template variables within workflow actions, enabling more dynamic and context-aware outputs.

We are improving AI regionalization by collecting detailed performance and availability data across providers and regions to ensure faster, more reliable model routing, including evaluation of platforms such as Groq, AWS Bedrock, and Azure AI.

We are introducing a unified application performance monitoring (APM) solution that brings together infrastructure and application monitoring into a single view, making it easier to understand system health and quickly identify and resolve issues.

Give faculty and admin an assistant that can pull grades, engagement data, and course info directly from the LMS.

Allow the assistant to chain multiple tools and refine its answers across steps, instead of responding in a single pass.

Enable teachers to customize their course assistant with safe, low-effort options while maintaining institutional-level controls for consistency and compliance.

Delegated management of sub-organizations

Add granular permissions so institutions can control which admin users can edit which parts of an assistant.

Track who changed what in an assistant's configuration and when.

Move UK customers to a dedicated regional server for data residency and performance.

Tracking, enforcement, controls, and UX for usage limits — the foundation for freemium → paid conversion, with AI Ops Assistant as the primary monetized surface. Target: live by July 2026.

Use existing LMS and ticketing/CRM integrations as additional sources of external user data, beyond what the injector signs and passes.

Allow students to converse with content from any course they have ever been enrolled in throughout their academic career, not just their current course.

Create a reusable library of workflow and custom instruction patterns that can be inserted into any assistant.

Provide pre-built assistant templates so institutions don't have to start from scratch.

Add new H5P content types to cover more learning styles and self-study use cases.

Ability for student generate feedback before submitting their assignment

Make AI-driven learning activities — roleplay, simulation, practice interviews, case work — placeable inside LMS courses like assignments: teachers configure the scenario and end condition, students complete it, grades and feedback flow back.

Create a standardized framework for building new integrations with institutional systems (SIS, CRM, library, etc.).

Connect the agent to CRM/SIS systems like Slate to read enrollment data and (later) trigger actions, with a reusable framework for adding more systems.

Bring feedback into chat with file support, event streaming, and basic student mode.

Post-March 23rd release: online learning, automatic settings selection, enhanced student feedback, plan caching, and controller improvements.

Let teachers control when and how AI-generated feedback is delivered and synced, instead of it being admin-only.

Display knowledge base images inline in responses instead of linking out.

Implement inline, clickable citations embedded within AI-generated responses, allowing users to view source material seamlessly as part of the answer flow

Play knowledge base videos directly in chat responses instead of linking out.

Let students attach files (documents, code, essays) directly in the conversation.

Cron / scheduling layer that runs agent workflows automatically in the background. Sits on top of the Sandbox / Execution Environment.

Send notifications via SMS, email, or WhatsApp — not just in-app.

Secure sandboxed environment where agents can run tasks in a real file system / compute context. Foundational, reusable capability that unlocks analysis, reports, and complex operations across features.

We will integrate fully configurable CSAT surveys into workflows, allowing organizations to define when, how often, and under what conditions feedback is collected.

Currently under consideration.

Expose insights data through an API so institutions can pull it into their own reporting tools.

Conversational querying of internal data, shipped as a tool inside Agentic Loop (not a standalone product). Main complexity is permissions, not the chat itself.

Currently under consideration.