📌 Release Overview

• 2026.20

• Unique hosted: May 18, 2026

• Self hosted: TBD

Chat Platform

🚀 What’s New (Business/End-User Focus)

User EXPERIMENTAL Agent Skills: All behind the feature flag FEATURE_FLAG_ENABLE_SKILLS_UN_20042, you can now trigger several Skills in a single prompt by combining / commands (e.g. /review-contract and /draft-email to my team lead). The assistant runs each Skill in sequence on the same context (including uploaded files) and shows the steps it took, so multi-step tasks can be handled in one turn instead of multiple back-and-forth messages. A typeahead picker lists available Skills as you type /.

Admin EXPERIMENTAL Add Skills to a Space: All behind the feature flag FEATURE_FLAG_ENABLE_SKILLS_UN_20042, enable the new Skills capability on a Space and pick a root folder; all skills inside it (and sub-folders) become available as / commands in that Space.

User Code Execution Card Font: Fixed CodeExecutionCard to render with the Inter font, matching the design specification.

User File Link Icon Sizing: Fixed file download link icon rendering at the correct size instead of being forced to 16px by the parent Button component.

User Download Button Visibility: Made the download and "Open in side panel" buttons in HTML rendering always visible instead of requiring hover.

User Skip Excel Ingestion with Code Execution: When an assistant has the code execution tool enabled, uploaded Excel/CSV files are now automatically handled by the code interpreter instead of the RAG ingestion pipeline, unless an admin has explicitly configured an ingestion mode.

NOTE: Even without this change, code execution outputs do not include references, as the output is governed by code execution instructions — skipping ingestion avoids unnecessary processing.

User User MCP Connectors: The user MCP connectors is now available from all apps via the user settings in New Navigation Menu. Users can view available MCP servers and authenticate to servers of their choice to connect the chat app with these MCP servers.

NOTE: Since it comes with new navigation menu, the feature flag for new navigation menu must be turned on to use.

Admin Added Timezone Parameter to Analytics: Analytics report can now be generated in the user’s desired timezone. This applies to when specifying a date range to generate a report for and the date-time formatting in the generated report. Date ranges are now parsed in the user-desired timezone or UTC, as it was (default), when not specified. Date-times in generated reports are formatted according to the user-desired timezone or UTC, as it was (default), when not specified.

The analytics front end now includes an interface to allow users to select desired timezones to generate a report in. This interface sits behind a FEATURE_FLAG_ANALYTICS_TIMEZONE_UN_18143 flag.

User Today's Briefing & Suggested Prompts on Chat Welcome View: When an assistant is configured with a briefing narrative and/or suggested prompts attached to the briefing, the chat welcome view now shows a briefing section, a rich-text/markdown summary that visually stands out from regular chat messages, together with a set of clickable prompt cards. Selecting a card prefills the chat input with the prompt, so users can kick off a conversation in one click. Long briefings collapse with a "Show more" toggle, and large prompt sets paginate into a carousel. If an assistant has no briefing or suggested prompts (or the flag is off), the standard welcome view is unchanged.

This feature is available in the new chat UI only and is disabled by default. It can be enabled upon request via feature flag. Briefings can be created, updated, or removed via the public API (PUT/GET/DELETE /public/briefings/{assistantId}).

User BETA New Side Navigation: A redesigned side navigation is now available across Chat, Knowledge Base, Manage Spaces, MCP Hub, and Settings. The shape is similar, but the experience differs from today's nav in a few important ways:

• Context-aware rail. A context panel next to the rail shows the right content for the active section — your chats and pinned spaces under Chat, the spaces list under Manage Spaces, MCP servers under MCP Hub, and an alphabetised list of admin tools (Analytics, Apps, Benchmarking, Cost Management, Feedback, Theme Editor, Access / User Management, AI Module Templates, Banner Notification, SCIM Keys) under Settings. The panel can be collapsed and re-opened from the close button on the rail.

• Inline space management. The spaces sidebar now has a per-row action menu for Pin / Unpin, mirroring the spaces landing page. Previously each of those required opening the landing page first. Pinned rows show a pin indicator that swaps to the action menu on hover, so the pin affordance and the menu coexist.

• Unified user / account menu. Signing out, switching theme, managing connectors, and other account actions now live in a single account menu in the rail footer instead of being split between the top header and the secondary menu.

• Alphabetised admin tools. The Settings panel lists admin tools (AI Module Templates, Analytics, Apps, Banner Notification, Benchmarking, Cost Management, Feedback, SCIM Keys, Theme Editor, Access / User Management) in a single alphabetised list, gated by each user's permissions, easier to scan than the per-area groupings today.

This feature is disabled by default. It can be enabled upon request via feature flag. When the flag is off, the existing top-bar navigation is rendered unchanged.

🎨 UI / UX Refresh of Chat Interface GA

User Chat Welcome View:

• We have introduced a warmer, more inviting entry: “How can I help you?” with a short description.

• The large, always-on upload area is removed; upload is available via drag-and-drop into the chat box or via a less prominent control, so the main focus is on asking a question.

• This new UI is not enabled by default (but can be enabled via the corresponding feature flag mentioned in the Infrastructure Changes) as it part of the UI refresh we are working on.

User Buttons inside the chat box: Chat input actions (upload, voice, tools, send) are moved inside the chat input area for a cleaner, more consistent look. Upload and voice are de‑emphasized so they’re not confused with the primary send action, while still easy to find when needed.

User Simpler prompt menu: My Prompts and Suggested Prompts are in a compact bar below the input with clear labels, so you can open and pick prompts faster.

User Easier tool selection: Tool selection is available via a “+” menu, with grouped tools (MCP tools) in dropdown groups for easier separation of concerns.

Chat Messages UI Refresh GA

User Cleaner answer layout: User messages use a bubble style on the right; assistant messages retain a flat style on the left for a clear question-and-answer distinction. Assistant message avatars are no longer shown, making threads easier to read, especially on mobile.

User Sources and content with the answer: Files and sources linked to an answer are shown as clickable badges directly under that answer’s text instead of as separate blocks between messages.

User Unified message actions: Copy, feedback (thumbs up/down), references, and debug sit in a single action row for assistant messages. References and debug open in a side panel so you can keep reading the thread while viewing references or debug info. On user messages, actions appear on hover to keep the default view minimal.

User Tables with copy: Tables in assistant messages use an updated design with a copy button in the lower-right of the table (visible on hover on desktop, always visible on mobile) so you can copy table data quickly.

User Updated suggested follow-up questions: Follow-up questions use a clearer design with a list-style layout so suggested next questions are easier to scan and tap.

User Floating hallucination indicator: The previous Answer Assessment button is removed. If no hallucination is detected, nothing is shown. When the model flags potential inaccuracies (moderate or severe hallucination), a floating button appears in the lower right (yellow for moderate, red for severe). Hover over the button to view the reason for the hallucination. It stays visible while you scroll so you can still see and act on the warning.

🔧 Improvements

User HTML Rendering Height Handling: Rewrote the iframe height-reporting mechanism to fix infinite height expansion, improve side panel rendering with native scrolling, and correctly measure content that uses body margins or percentage-based heights.

Admin MCP Server Error Messages: Improved error handling when connecting to MCP servers — now shows user-friendly messages for network-level errors (DNS failures, timeouts, connection refused) and includes HTTP status codes for non-2xx responses.

RAG

🚀 What’s New (Business/End-User Focus)

🔧 Improvements

User Performance improvements for Search: Reduced unnecessary data processing during search, so the service spends less time parsing and returning large results.

🐞 Bug Fixes

User Chunking Strategies: Fixed a bug where pdfs do not respect the selected chunking strategy in the set ingestion config. PDFs are now processed using the selected chunking strategies (like CUSTOM_CHUNKING_API) providing consistent ingestion behaviour.

Agentic Table

🚀 What’s New (Business/End-User Focus)

User Verbatim question import: Agentic Table now supports direct CSV/XLSX question-file import. Space admins can configure question_extraction.mode as strict_verbatim to import matching rows directly, or prefilter_verbatim to try direct import first and fall back to LLM extraction when validation fails. Existing spaces continue to use the default extract mode.

🔧 Improvements

User Clearer handling of expired files in the Agentic Table: Files that have passed the retention period are now visually marked as expired in the "View Files" dropdown of an Agentic Table.

ADMIN Cleaner question details in Word export: Question headings in the Word export now contain only the clean question text instead of an inline raw HTML "details" block rendered as a huge H2 heading. Section, question type, answer options and parent context now appear under a tidy "Details:" sub-heading, and can be turned off entirely via the new include_question_details_in_export flag in docx_generator_config field of the config.

🐞 Bug Fixes

User Word Export Title Population: fixed title population in Word (.docx) report export of an agentic table sheet.

User Agentic Table only showing 20 rows on creation: Fixed an issue where creating an Agentic Table from a large dataset (200+ rows) only displayed the first 20 rows until the page was reloaded. All rows are now visible immediately after table creation, without requiring a reload.

User Same file used as both question and source: Fixed an issue where a file uploaded as a question file in an Agentic Table was silently excluded from search even when the user explicitly selected it as a source, leading to empty or source-less answers. The user's explicit source selection now wins, so a shared question/source file is correctly used as a search source.

User Word export stuck on "Preparing your file": Fixed an issue where exporting a sheet as a Word document could hang on "Preparing your file" when questions or section headers contained hidden control characters. These characters are now sanitised before export, so Word exports complete reliably.

Unique AI

🚀 What’s New (Business/End-User Focus)

Admin Persistent File Attachments for Code Execution: You can now configure Code Execution to always upload specific files into the container by referencing their content IDs. This is useful for attaching reusable assets — for example, ensuring a standard template file is automatically available in every execution environment. The configuration can be found in Code Execution--> Tool--> Additional Uploaded Documents

Admin BETA Dynamic Metadata Filtering in Space 2.0: Internal Search in Space 2.0 now supports dynamic metadata filtering based on parameters extracted directly from user queries. This feature enables more context-aware and precise retrieval behavior while keeping filtering logic fully configurable through the Knowledge Scope UI. More details and how to use dynamic filters can be found here. This feature is currently BETA and requires enabling the feature flag FEATURE_FLAG_ENABLE_DYNAMIC_SCOPE_RULE_UN_17614.

🐞 Bug Fixes

User Improved Uploaded File Search with Exclusive Internal Search Tools: Fixed an issue where enabling an exclusive internal search tool prevented uploaded files from being included in search results. Uploaded file search now remains available independently, ensuring queries can search both internal sources and conversation-uploaded files as expected.

User Fixed Orphaned “Triggered Tool Calls” Entries in Steps Timeline: Fixed an issue where the Steps timeline could show an extra “Triggered Tool Calls” entry without a matching follow-up step after multiple Internal Search runs in the same assistant turn. Subsequent searches now correctly create their own timeline entries instead of reusing and overwriting the previous Internal Search log.

Language Model

Admin GPT-5.4 and GPT-5.5 Now Available as DataZoneStandard via Azure: GPT-5.4 and GPT-5.5 are now available as DataZoneStandard deployments via Azure (processing restriction to EU). The corresponding enums areAZURE_GPT_54_2026_0305 and AZURE_GPT_55_2026_0424. Both models are supported by UniqueAI and appear in the model dropdown if they are deployed.