Q1 2026: AI-Powered Authoring, Call Recording, and a Sharper Digital Twin Editor
- Gilad Tzori
- 1 day ago
- 6 min read
Q1 2026 was the quarter the frontline.io AI assistant stopped being a search tool and started being a co-author. Remote Support gained the recording and compliance features that enterprise teams have been waiting for. And across the authoring stack — Digital Twin Editor, Toolbox, Immersive Room — the kind of precision improvements that don’t make headlines but make every day measurably better. Here’s what shipped.
AI Platform
Until now, the frontline AI assistant was a smart reader — it could find information, answer questions, and surface relevant content. What it couldn’t do was act. That changes in Q1. The assistant can now edit procedures directly, apply changes without a manual review step, and give administrators full visibility and control over how AI is used across the workspace. It’s a meaningful shift: from a tool that tells you what to do, to one that helps you do it.
Model Selection
Users can now choose which AI model backs their chat session. The model selector supports four options: Frontline AI (a balanced, recommended default), ChatGPT 5 Mini (highest capability for complex queries), ChatGPT 5.2 Chat (optimized for speed), and ChatGPT 4o Mini (lightweight, low-latency). The selection persists across sessions per user, so teams that need fast answers and teams that need deep reasoning can each pick the model that fits their workflow.
Auto-Apply Changes
Previously, every AI-suggested edit required a manual approval step. With Auto-Apply, the system now executes content modifications — procedure text, metadata, media references — immediately upon request and surfaces a structured summary of what changed. This significantly reduces the edit-review cycle for content creators working across large procedure libraries.
Real-Time Thinking Status
When the AI takes longer than one second to respond, the chat interface now displays contextual status indicators — “Thinking” for general responses, “Searching” when querying the knowledge base, and other processing-specific labels. This eliminates the “is it working?” uncertainty, especially during complex multi-step queries.
AI Studio: Usage Dashboard & Token Limits
Workspace administrators now see a token usage bar on the AI Studio page showing tokens used, with automatic visual warnings at 80% and 100% thresholds. On the back-office side, a new Token Limit configuration lets admins set a hard ceiling on total token consumption per workspace — a critical control for organizations managing AI costs across multiple business units.
Knowledge Source: Sub-Workspace Exposure Control
Enterprises with hierarchical workspace structures can now control AI knowledge sharing at the file level. A new “Sub-workspace permitted” column in the AI Studio files table lets managers explicitly allow or deny sub-workspace access to each knowledge file, inheriting the parent workspace’s role and exposure settings when permitted. This addresses a common governance concern in multi-site deployments where certain documentation should remain scoped to a specific facility or division.
PDF Parsing for Knowledge Base
We improved how the AI knowledge base ingests and indexes PDF documents, resolving parsing issues that had affected search accuracy and RAG-based retrieval. Teams uploading technical manuals, schematics, and compliance documentation as PDFs will see more reliable AI responses grounded in those sources.
AI Studio UX Improvements
Beyond the headline features, the team resolved dozens of issues across AI Studio: file rename handling, batch actions for role assignment and enable/disable operations, system feedback toasts for upload, delete, and publish actions, file reference resolution in project-scoped chats, and a 404 access bug. The net result is a substantially more reliable content management experience within AI Studio.
Remote Support Web
Real-time video and AR annotation have always been the core of Remote Assist. But enterprise customers need more than a great call — they need a record of it, a way to measure it, and controls that satisfy their compliance teams. The 26.2 Web release delivers all three.
Call Recording
Remote Support Engineers and Managers can now record audio and video of live sessions. The implementation provides manual start/stop controls, clear on-screen recording indicators visible to all participants, and automatic save-on-disconnect to prevent data loss. Recordings are stored alongside the session log and are accessible for playback and download from the session record view. The feature is gated by a per-workspace feature flag in the Backoffice, giving IT administrators control over where recording is enabled.
Post-Call Survey
Workspace administrators can define up to five custom survey questions that are automatically presented to participants at the end of a remote support call. Responses are linked to the session record and aggregated in a new Remote Support tab within workspace management, providing native CSAT and quality metrics without relying on third-party survey tools.
Guest Terms & Conditions
Guest users joining a remote support session via the web are now presented with a terms-of-use checkbox before they can connect. The checkbox is checked by default and includes a direct link to the full terms document. This addition addresses compliance requirements for organizations that need explicit consent documentation for external participant interactions.
Guest Session Stability
We resolved an issue where guest-initiated sessions could produce a broken remote support state. The fix ensures clean session handling for web-based guest participants.
Digital Twin Editor
The Digital Twin Editor received precision and organisation improvements aimed at content creators managing complex 3D environments.
Multi-Angle Reference Grid
The editor’s reference grid now automatically rotates to face the camera when switching to orthographic side views — front, back, left, and right. This behavior mirrors what professional 3D authoring tools like Blender and Maya provide, where the grid always stays visible as a spatial reference behind the model rather than disappearing when viewed edge-on. The grid aligns to the XY plane for front/back views and the YZ plane for left/right views, snapping back to the XZ ground plane in perspective mode.
Configuration Folder System (Phase 2)
Content creators working with large numbers of configurations can now organise them into folders. The system supports full CRUD operations — create, rename, delete — and drag-and-drop for moving configurations between folders. Selecting a folder selects all configurations within it, enabling bulk operations that were previously tedious in flat-list views.
Transform Component Accuracy
We corrected an issue where position and rotation values displayed incorrect data after certain editing operations, ensuring the transform component now accurately reflects the object’s state at all times.
Toolbox & Tool Editor
The Toolbox and Tool Editor received a set of usability improvements focused on reducing friction for content creators managing tool libraries.
Tool Library Multi-Selection
Users can now select and operate on multiple tools simultaneously in the Tool Library, eliminating repetitive one-by-one workflows when reorganising or bulk-editing tool sets.
Drag-to-Change Category
Tools can be reassigned to different categories by dragging them directly, replacing the previous multi-step reclassification process.
Rename from Library & Refresh
Tools can now be renamed inline from the library view, and the tool list can be refreshed without reloading the entire editor.
Tool Editor Fixes
We corrected pivot editing behavior where position and rotation values were incorrect post-edit, resolved root part visibility issues, and fixed category assignment on editor launch.
Immersive Room
Hosts in the Immersive Room can now access and open procedures that are connected to the loaded digital twin. Previously, entering an immersive session with a digital twin did not surface its associated procedures — requiring users to exit, find the procedure, and re-enter. The workflow is now seamless: enter the room, see the procedure list, and launch directly.
AR / XR Runtime
QR Target Anchoring
Digital twins can now be anchored to QR targets with a configurable offset, allowing precise placement relative to a physical marker. We also resolved an issue where loading a digital twin via QR target could twist the environment orientation, and fixed beam interaction so operators can click and operate targets reliably in MR.
Object Target Lock Frame
Fixed an issue preventing users from adjusting the lock frame on AR object targets — a critical control for aligning digital overlays with physical equipment in field scenarios.
Infrastructure
On-Premises Fixes
A targeted set of fixes for on-premises deployments addresses issues that accumulated as cloud-first features were backported: 3D procedure animations now play correctly, the Interactive Flow engine properly connects animations to flow nodes, the Immersive Room loads digital twin and environment dropdowns reliably, 3D Scene bookmarks save correctly, and the Global Material Library is accessible in the Digital Twin Editor. These fixes are consolidated into the upcoming 26.1 On-Prem release.
Salesforce Remote Support Integration
The Salesforce integration for Remote Support — enabling session initiation and logging directly from Salesforce Work Orders — was released to production. This is a key enablement for enterprise customers running their field service operations through Salesforce.
Looking Ahead
The through-line across Q1 is the same whether you’re looking at AI, Remote Support, or the authoring tools: the platform is maturing from capable to enterprise-ready. Governance controls. Audit trails. Compliance features. Precision where precision matters.
Q2 continues that trajectory. If you’re evaluating frontline.io for a specific use case, or want to understand how any of these updates apply to your deployment — we’re happy to talk.
