You can also browse, search and filter all release notes directly in the application → Open release notes in TAI.
Company admins now get active-user & session KPIs on their dashboard
New Feature. Only Company Admins.
The Dashboard and Analytics now surface headline KPIs for your company: active users and sessions, each over the last 24 hours / 7 days / 28 days with the percentage change versus the previous period.
Active users counts people who actually ran an agent; sessions includes report runs. A dash shows when there's no prior-period activity to compare against.
See what your agent is doing — a live activity timeline replaces the silent wait
New Feature.
While an agent works on your request, the chat now shows a live timeline of what it is doing — thinking, searching the knowledge base, querying your data, searching the web, working with files, or running subtasks. Completed steps get a checkmark, the current step pulses, and the timeline collapses into a compact summary once the answer starts streaming. Reopening an older conversation shows the same summary for past replies.
Company admins can now see each connector's billing tier at a glance
Improvement. Only Company Admins.
Every connector belongs to a billing tier — Included, Default, or Premium — that decides how its calls are priced. That tier is now shown as a badge on the connector list and on each connector's detail page, so you can tell at a glance which connectors are included in your plan and which bill at a premium. Display only for now; the tier itself is set by Nezasa.
Company admins can now show an agent's thinking to end-users — compact or full
Improvement. Only Company Admins.
An agent's Model settings now control its reasoning display the same way for everyone, not just Nezasa staff: pick None to hide it, Compact for a condensed activity timeline, or Verbose for the full thought process — so end-users can finally follow along with how an agent reaches its answer. A separate Thinking switch turns the model's reasoning on or off (off is faster and cheaper on simple agents), and a Narrator toggle controls the live one-line summary of what the agent is working on.
Saving an agent now keeps you on the settings page
Improvement.
Clicking Save changes on an agent's settings no longer bounces you back to the agents list. The page stays put, a confirmation toast tells you the save succeeded, and your edits remain on screen — so you can keep refining the same agent without reopening it.
During long thinking, the chat now tells you what it's actually working on
Improvement.
When an agent spends a while thinking, the activity timeline used to show a generic 'Thinking…' that cycled through filler. Now it shows a live, plain-language line describing what the assistant is working on right now — refreshed every few seconds — so a long step feels like real progress instead of a stall.
Report PDF exports are back to the clean, polished layout
Fix.
When a report exported as a plain-text PDF — cramped, unstyled, and sometimes with the agent's working notes left in — the export now renders in the polished report layout instead: a clear title, spaced sections, and properly styled tables, ready to share with a customer without rework.
The agent's analysis chatter ("let me load the skill…") is stripped from the exported document, so only the report itself reaches the page.
Reports that already exported cleanly are unchanged.