What's new

Short entries about changes visible to users. Newest first. Entries are added by the whats-new skill; at release time the "Unreleased" section becomes a version heading.

Unreleased

v0.11.0 — 2026-06-11

English conversations now get fully English assistance

When the conversation language is set to English, every AI surface now answers in English: the conversation Map, Story recommendations, agenda check-off reasons and suggested questions — not just hints and Pulse as before. Each AI prompt now has a native English version instead of a Polish prompt with a "respond in English" note, which makes English output noticeably more natural. Switching the language mid-meeting applies to the very next AI pass. This covers the browser (web) version too.

English interface by default

On a fresh install with no saved preference, the app interface now starts in English (Polish systems still get Polish automatically). Your saved language choice is unaffected — you can switch between English and Polish in Settings or in the assist window's quick settings at any time. The web app now also boots in your saved interface language instead of always starting in Polish.

Choose the AI model for agenda tracking

The "AI models per task" section in Settings (and the quick-settings popover in the assist window) gained an "Agenda check-off" entry — you can now pin a specific AI model for live agenda tracking, like for every other task.

See the guide: Choosing AI models per task

Quick settings right inside the assist window

The settings (gear) icon in the assist window now opens a quick-settings popover instead of pulling up the main window. Without interrupting the meeting you can switch the conversation language, the interface language, the AI model for each task (Map, Story, Pulse…) and the microphone — changes apply immediately. "Open full settings" at the bottom still takes you to the full Settings screen when you need more.

See the guide: Changing settings during a meeting

Microphone selection now works — and switches live

The microphone picked in Settings (and in the new quick-settings popover) is now actually used for capturing your voice — previously the app always used the system default regardless of the setting. Changing the microphone during a meeting takes effect right away, with under a second of pause in transcription. If the chosen device is unplugged, the app falls back to the system default instead of going silent.

More breathing room under the top bar

Screens in the main window no longer start right at the edge of the top bar (organization and account) — content now has clear spacing below it on every screen.

Online documentation

Guides and this changelog are now published at docs.joltuar.app. The site rebuilds automatically whenever the documentation changes.

Account Settings redesigned with tabs (#43)

The Account Settings screen was rebuilt. Settings are now grouped into tabs at the top — Account, Assistant, Audio & transcription, Plan & usage, Privacy and About — and a search box at the top jumps straight to the right section. Every card and control got a cleaner, refreshed look and works in both light and dark mode. AI models — including the per-task picker — now live under the "Assistant" tab.

See the guide: Account Settings in a tabbed layout

Pulse speaks one language (commit 3fec930)

Signal summaries on the Pulse tab now always follow the language selected for the conversation (Polish or English) — no more mix of Polish and English hints in a single view. Signal labels (Objection, Question, etc.) intentionally stay in English — they are fixed category markers.

v0.10.0 — 2026-06-10

Choosing AI models per task (commit 8499b75)

Account Settings gained an "AI models per task" section. You can separately choose which AI model handles Map, Story and Pulse points, Pulse hints and suggested questions in Map — or leave "Auto (default)" to let the app pick a fast model for point extraction and the global one for everything else. The global model list also gained 20 leaderboard models (commit 14ffc29).

See the guide: Choosing AI models per task