This page lists recent changes to SUPERWISE Chat — new features, improvements, and fixes. Entries are grouped by release, newest first. Most changes roll out automatically; you do not need to update or reinstall anything.
Connect a Confluence space so its pages become searchable knowledge your assistants can cite — without copying files by hand. Every connected document is virus- and sensitive-data scanned on the way in. See Upload a file.
Web crawl as a background job
Crawling a website into your knowledge now runs in the background, so large sites import reliably without tying up your conversation. Progress shows inline while it runs.
Notes
A dedicated place to keep written notes and save useful AI responses. Notes carry a title, category, and a link back to where they came from.
Channels and Direct Messages
Persistent multi-person spaces (Channels, with threads, pinned messages, and a moderator role) and private group chats (Direct Messages), with the assistant available in both.
Shared, real-time conversations
Share a conversation with teammates and work in it together live — everyone sees messages stream in, and each person’s messages are clearly attributed to them.
Conversation export
Export any conversation as PDF, Markdown, plain text, or JSON. See Export.
Your organization’s data is now isolated at the database layer itself, so one organization can never see another’s conversations, knowledge, or settings. See Tenant isolation.
Clearer governance details on every answer
Each answer carries a trust indicator you can open to see how it was produced — the steps taken, sources used, and the safety checks that ran. See Why it said no.
Better citations
Answers grounded in your knowledge show numbered, clickable source cards with a snippet and the access scope. When no matching source is found, the answer tells you it is “answering from general knowledge.”
Coreference and clarification
The assistant follows pronouns and references across a conversation more reliably, and asks a focused clarifying question when your request is ambiguous rather than guessing.
Multi-step workflows that combine AI actions, external calls, timed waits, and human approval steps. Approval requests arrive in your notifications and run on a durable engine, so long-running and waiting steps survive interruptions.
Single sign-on (SSO)
SSO sign-in is available on Business plans and above. See Identity and SSO.
Group related conversations with a shared assistant, knowledge, and settings. New conversations created inside a folder join it automatically. See Organize conversations.
History, pin, rename, archive, search
Rename and pin conversations, archive (and restore) them, and search within a conversation with Ctrl/⌘+F.
Response actions
One-click follow-ups under each answer — Explain more, Refine this, Try a different approach, Compare (side-by-side variants), plus copy and regenerate.
Curated, searchable document collections the assistant consults and cites, on a scope ladder from a single chat up to your whole organization.
File upload
Upload PDF, Office (DOCX/XLSX/PPTX), CSV, text, Markdown, code, and images (up to 100 MB per file). Files are scanned before they become usable. See Upload a file.
Web search
Let an answer pull from the live web when you need current information, with sources cited inline. See Web search.
Document references (#)
Type # in the composer to attach a specific knowledge document to a single question.
Reusable configurations — system prompt, persona, linked knowledge, and guardrails — built through a conversational builder. See Choosing an assistant.
Interaction tiers
Choose how deeply the assistant thinks per question: Interactive (fastest), Thinking (default), Research, and Thought Partner. See Tiers explained.
Memory
Personal memory remembers your preferences and facts (private to you, reviewed before it is saved); organization memories are shared by your administrator.
Personas and preferences
Set tone, response length, theme, and send behavior. Custom personas are available where enabled. See Preferences.
