Nojoin User Guide

This guide covers day-to-day use after Nojoin has been deployed and your account has been created.

For deployment, administration, calendar provider setup, backup operations, and detailed capture troubleshooting, use the dedicated guides in the docs folder.

First Run

1. Open Nojoin in Chrome on Windows, Linux, or macOS for shared-audio recording, another supported desktop Chromium browser, or Chrome on Android/iOS for microphone-only recording.
2. Sign in with your account.
3. Open Settings > Capture if you need to choose a microphone or adjust shared-audio and microphone gain.
4. Use the Meet Now card on the dashboard and click Start Meeting to create a short test recording.
5. In the browser share picker, choose the meeting tab, window, or screen and enable the browser’s audio-sharing or system-audio option when it is offered.
6. Speak briefly and confirm the live waveform responds. If AI is configured, Meeting Edge guidance may appear once enough speech accumulates.
7. Stop the recording, open it in the /recordings workspace, and wait for processing to complete.

Firefox, Safari, and mobile browsers other than Chrome can review recordings but cannot start live capture. Chrome on macOS is supported for recording; other Chromium-family browsers on macOS are best-effort. See CAPTURE.md for the full support matrix.

Tours And Onboarding

Nojoin includes guided tours for first-time users.

• The dashboard tour introduces navigation, recording, importing, capture setup, and settings.
• The transcript tour introduces the recording detail view when a recording is opened for the first time.
• Tours can be restarted later from the Help settings area.

Dashboard

The root route is the operational home surface for Nojoin.

It brings together:

• Meet Now card for live browser capture.
• Calendar context through month and agenda views, including recorded meeting history.
• Task List for quick personal follow-up capture.

On desktop viewports around 1920x1080 and smaller, Nojoin automatically shifts into a denser desktop layout so more dashboard, recordings, transcript, notes, and settings content remains visible without affecting the roomier large-monitor layout.

Calendar Surface

• Switch between month and agenda views.
• Use Today to jump back to the current date context.
• In month view, selecting a day opens a day agenda, and selecting today shows a live now marker against timed events.
• Unlinked Nojoin recordings appear on the dashboard calendar as orange meeting items, while Google or Microsoft calendar sources keep their own colours.
• Recorded meeting cards surface tags, speakers, and timestamps directly inside the selected-day agenda.
• The agenda view is month-scoped and includes both synced calendar events and unlinked Nojoin meeting history for the viewed month.
• In the current month, the agenda starts at the next upcoming item; use Show past events to expand earlier items. Past months always show their full history.
• Clicking a timed event bubble in the day timeline opens a details popover with the full title, time, calendar, location, join link, and linked recordings.
• Live events with a trusted meeting link show a Join button directly on the timeline bubble.
• Meeting links render as compact Join meeting labels with the provider host rather than raw URLs; the full URL appears on hover.
• Events continuing past midnight show dashed edges and chevrons on the day timeline.
• Event times render in your configured Nojoin timezone.

Read CALENDAR.md for connection and setup details.

Task List

The Task List is a personal dashboard list for quick follow-up work.

You can create, rename, complete, reopen, archive, delete, and schedule tasks. Active deadlines show a live time-remaining badge. Archived tasks disappear from the dashboard immediately, including tasks that were already completed.

Tasks Workspace

The dedicated Tasks page sits in the main navigation between Dashboard and People.

Use it to manage tasks more holistically:

• Select Create Task to open the task creation form when you need to add a richer task.
• Open shows active, non-archived tasks.
• Completed shows finished, non-archived tasks that can still be reopened.
• Archived shows hidden tasks and lets you restore them to the active task surfaces.
• Task cards support a title, body, deadline, permanent delete, direct recording links, and the same recording tag taxonomy used elsewhere in Nojoin.

Delete remains permanent. Use archive when you want to hide a task without losing it.

People Workspace

The dedicated People page sits in the main navigation between Tasks and Recordings.

Use it to manage your shared speaker library:

• Review people records and stored voiceprints in one place.
• Search and filter the library when you need to find a known speaker quickly.
• Organise people with reusable people tags.
• Open batch editing and cleanup flows for broader speaker-library maintenance.
• Delete a person to remove them from the library. Deletion demotes rather than erases: the shared record, its voiceprint, and its contact details are removed, while each of their meetings keeps the speaker under the same name as a recording-local speaker.

Live Recording

Live recording is browser-native. On supported desktop Chromium browsers, Nojoin captures shared tab/window/screen audio plus microphone audio from the web app. On Chrome for Android and iOS, Nojoin records the phone microphone only.

1. Open the dashboard.
2. In the Meet Now card, click Start Meeting.
3. On desktop, select a meeting tab, application window, or entire screen in the browser share picker.
4. On desktop, enable the browser’s audio-sharing or system-audio option when it is offered before selecting Share.
5. Allow microphone access if prompted.
6. On mobile Chrome, keep the phone close enough for the microphone to hear the meeting audio.
7. Keep the Nojoin tab open and the device awake until the meeting ends.

Tab sharing is usually the best choice for browser-based meetings because it most reliably exposes meeting audio. Window and screen sharing can work, but audio availability depends on browser and operating-system support.

If you close the browser share picker with Cancel, Nojoin silently returns to the pre-start state and no recording begins.

Mobile Chrome does not capture meeting tab, app, headset, or system audio. It is useful for microphone or in-room speakerphone capture only.

While recording, Nojoin shows recording state, duration, upload state, a live waveform, Meeting Edge guidance, your live notes panel, and collapsed processing visibility.

On mobile and narrow tablet layouts, Nojoin uses compact navigation with a menu button on the main dashboard surfaces. The active recording workspace and processed recording detail view both provide a native back control so you can return to the recordings list without relying on the browser’s history buttons. Mobile Meeting Chat also includes its own back action to return to the meeting workspace.

You can switch to another browser tab, window, or application while recording. Nojoin only pauses automatically when the Nojoin tab is refreshed, closed, or navigated away from the active recording.

While a recording is active, a floating badge appears at the top-centre of the viewport on every page. The badge shows the recording status, elapsed time, and pause, resume, and stop controls. Clicking the badge navigates to the recording detail page. You can control the recording from any page without navigating back to the recording workspace first.

Pause, Resume, Stop, And Discard

• Pause temporarily stops capture while preserving uploaded segments.
• Resume opens the browser share picker again and continues the same recording.
• Stop finalises the recording and starts processing.
• Discard permanently removes an in-progress recording in one step. It stops capture, cancels any processing, deletes the captured audio, and removes the meeting. Nojoin asks you to confirm first because this cannot be undone.

Discard is available from the live recording controls, the floating recording badge, the resume-or-discard modal, and the recordings menu, so you can abandon a recording from wherever you are.

If the browser is closed, refreshed, or loses the active recording page during capture, Nojoin pauses the recording to protect already uploaded data. When you return, Nojoin requires you to resume or discard that recording before starting anything else.

Paused recordings are retained indefinitely until you resume or discard them.

Live Transcription

Nojoin still runs a live transcription lane during capture, but the recording page no longer shows provisional live transcript text while a meeting is in flight.

That live lane now works in the background to support Meeting Edge and to speed up later processing. The user-facing transcript appears after the recording has been finalized and the authoritative processing pass has produced review-ready output.

Meeting Edge

Meeting Edge uses the recent live transcript window, an internally maintained rolling summary of the meeting so far (decisions, open threads, and action items), its own previous suggestions (so guidance stays fresh instead of repeating), your optional focus text, your manual notes, and linked calendar context when available.

It can surface live questions, missed points, and quick concept help during a meeting. In Settings > AI > Meeting Edge, everyone can tune the per-user Technical context slider to make concept explanations stricter or more detailed; administrators can additionally enable or disable Meeting Edge install-wide and choose a separate Meeting Edge model (if left empty, Nojoin reuses the main model).

Importing Recordings

You can import existing audio files directly through the web client.

Supported formats include WAV, MP3, M4A, AAC, WebM, OGG, FLAC, MP4, WMA, and OPUS.

The import flow validates the file, builds the canonical media artifacts, and queues background processing. Imports skip the live capture workflow but share the same final processing pipeline as live recordings.

Discard Recording

Use Discard Recording from the recordings list or recording actions when a meeting is still recording, paused, queued, or processing and you no longer want it.

Discard Recording:

• Revokes any running processing task for that meeting.
• Closes any active live upload or finalisation session.
• Deletes the captured audio and derived files.
• Removes the recording entirely, so there is no leftover Cancelled entry to clean up afterwards.

Because it permanently deletes the meeting, Nojoin asks you to confirm before discarding. To remove a meeting that has already finished processing, use Delete instead.

Retry Processing

If a recording fails or you want to rebuild the generated meeting artifacts, use Retry Processing.

Retry Processing clears transcript-derived generated state, preserves recording metadata, tags, uploaded documents, and user-authored notes, then records a fresh processing timing sample for future ETA calculations.

Reprocess A Recording

From the recording detail page you can choose Reprocess at higher quality. This re-runs the full pipeline after you change the transcription engine or model in Settings.

Reprocessing clears and rebuilds transcript and generated artifacts while preserving metadata, tags, documents, and user-authored notes. Older meetings that predate the unified pipeline cutover may require reprocess before transcript or speaker edits are available.

Transcript And Playback

Within a processed recording you can:

• Play the aligned web proxy.
• Follow synced transcript highlighting.
• Click transcript text to seek playback.
• Edit transcript text and speaker assignments.
• Export transcript-only, notes-only, or combined text output.

Historical recordings carried forward from before the unified pipeline cutover may open in a read-only compatibility state. Playback, transcript viewing, and export remain available, but transcript or speaker edits require explicit reprocess first.

Speaker Management

Nojoin maintains a global speaker library across recordings.

Common workflows include linking an unknown in-recording speaker to an existing global speaker, promoting a recording speaker into the People library, creating or updating voiceprints, recalibrating voiceprints from better samples, and merging duplicate speakers.

Inferred speaker names from final processing and manual retry flows are applied automatically to speakers that do not already have a trusted name. Manual speaker edits are authoritative: renaming, merging, promoting, or deleting a speaker always overrides inferred names, and speakers you have already named are never renamed by inference.

Notes, Chat, Documents, And Search

Processed recordings can include Markdown notes, AI-generated meeting notes, meeting chat, uploaded documents, and transcript/document search.

• Notes are stored with the recording and can be edited after processing.
• Generate Notes runs a notes-only AI pass when AI is configured.
• Meeting Chat answers questions from the transcript, notes, and linked documents.
• Documents can be uploaded to support meeting context and later search.
• Search spans recordings, transcript text, notes, tags, and document content where available.

Calendar Features

Nojoin can connect to Google Calendar and Microsoft Calendar when an administrator has configured provider credentials.

Calendar events can provide meeting context, dashboard agenda views, linked recording history, and Meeting Edge context. Read CALENDAR.md before changing calendar provider settings or troubleshooting OAuth.

AI Assistant Connections (MCP)

Nojoin includes a read-only MCP connector so AI assistants such as Claude can search your recordings and read transcripts, meeting notes, and tags on your behalf. Add https://your-nojoin-domain/mcp as a custom connector in the assistant and approve access on Nojoin’s authorisation page. Active connections are listed under Settings → Personal → Connected Apps, where each one can be revoked. See MCP.md for setup, supported clients, and troubleshooting.

Settings

Settings are grouped by task.

• Profile: account details and password changes.
• Capture: microphone selection, shared-audio gain, microphone gain, browser audio-processing toggles, and a local mic input test for browser recording.
• AI: per-user AI routing (the server’s configured provider, or your own Claude or ChatGPT subscription), the server provider and model, Meeting Edge, automatic meeting intelligence, language, and secondary-provider fallback. Install-wide controls (provider, models, the Ollama endpoint, fallback, and “Enable Meeting Edge”) are shown only to administrators; a non-admin sees a read-only summary of the active provider instead.
• Transcription: transcription backend and model choices.
• Calendar: user calendar connections and timezone behaviour.
• Help: tours and support surfaces.
• Admin: user, invitations, CLI usage, system, provider, release, and maintenance settings for administrators.

Language Preferences

Use Settings > AI > Language preferences to configure two independent choices:

• Transcription language controls ASR. The default is Auto-detect. Whisper supports auto-detection or a forced language, Canary supports the listed forced languages, and Parakeet continues to use multilingual auto-detection without forced-language support.
• Notes generation language controls AI-generated meeting titles, Markdown headings, summaries, detailed notes, and action items. The default is English. British English, American English, the transcription language, another listed language, or a custom language/style instruction can be selected.

The LLM prompt and machine-readable response contract remain stable even when generated content is localized: JSON keys, speaker labels, and application-owned metadata are not translated.

Changing the transcription language affects new transcription work. Use Reprocess at higher quality to rebuild an existing recording under a different transcription-language preference. Changing the notes language does not translate saved notes automatically; run Generate Notes or reprocess the meeting to generate new notes.

Language preferences are per-user. Per-meeting overrides, full interface translation, Meeting Edge/chat language controls, and speech translation are not part of this setting.

Your own subscription (CLI OAuth)

Instead of using the server’s configured provider, you can route your own AI through your own Claude (Pro/Max) or ChatGPT (Plus/Pro) subscription — usually faster, and you can pick a stronger model. In Settings > AI > AI routing, connect a subscription in the “Your AI subscription” panel, then set routing to My own AI subscription:

• Claude — open the grant link and paste back the code Anthropic shows you.
• ChatGPT — open the sign-in page, enter the code shown, and approve access; the panel finishes automatically once you approve.

Pick a model for notes/chat and, optionally, a faster live model for Meeting Edge. If you connect both, choose which one is active. When your subscription is unavailable or usage-limited, Nojoin falls back to the server’s default provider chain — its primary provider first, then its secondary if that also fails. Once you start using it, the panel shows your recent token usage.

Because this uses your subscription quota, a cheaper model conserves quota, not money. If you reach your limit, Nojoin shows a reset time and falls back to the server’s default provider chain — its primary provider first, then its secondary if that also fails. Disconnecting removes your stored credential.

CLI usage and quota (admin)

Administrators can review per-user Claude-subscription usage under Settings > Administration > CLI usage & quota: a searchable table of each user’s token usage over the last 7 days, last 30 days, and lifetime, alongside their rate-limit status. Tokens count what Nojoin sent through each user’s own subscription. A subscription exposes no live “remaining quota” figure, so the status column reflects the rate-limit signal (OK, approaching a limit, or limited until a reset time) rather than a balance. Usage is shown in tokens, never money — a subscription is flat-rate, so a per-turn currency figure would be misleading.

Secondary LLM Provider

Nojoin supports configuring a secondary LLM provider as a fallback. When the primary provider fails with any error, the system automatically retries the request with the secondary provider. This applies to all AI features: Meeting Edge, meeting intelligence, speaker inference, and meeting chat.

The secondary provider has its own independent configuration:

• Provider selection (Gemini, OpenAI, Anthropic, or Ollama).
• Model and live model choices.
• API key or Ollama URL.

Configure the secondary provider through environment variables prefixed with SECONDARY_ (e.g., SECONDARY_LLM_PROVIDER, SECONDARY_GEMINI_API_KEY). Leave SECONDARY_LLM_PROVIDER empty to disable fallback. The secondary provider configuration is visible in Settings > AI for administrators.

Troubleshooting

• If live capture is unavailable, switch to Chrome on desktop for shared-audio recording or Chrome on Android/iOS for microphone-only recording.
• If remote participants are missing, start again and enable shared audio in the browser picker.
• If the microphone is missing, grant microphone permission and check Settings > Capture.
• If Nojoin reports a paused recording, resume or discard it before starting another capture.
• If processing fails, use Retry Processing or check the administrator logs.
• If calendar sync fails, review provider setup in CALENDAR.md.

Updates

The Updates area shows the installed server version, latest available release, and release notes. Release metadata comes from GitHub Releases when the deployment can reach GitHub.