Getting Started

IntelliPrompter runs as a desktop app. You install it, sign in once, configure an AI provider, and you are ready to run your first meeting. This page walks through each step. Plan on 5 minutes start to finish.

• macOS 14.2 (Sonoma) or newer.
• Apple Silicon or Intel processor.
• 8 GB RAM minimum.
• A beta invite. Request access at intelliprompter.com/download.

1. From your invite email, download the IntelliPrompter DMG.
2. Open the DMG and drag the app to your /Applications folder.
3. Open IntelliPrompter from /Applications.
4. On first launch, macOS asks for microphone and screen recording permissions. Grant both. The app cannot transcribe without microphone access. Screen recording is required for system-audio capture in Full mode (so the app can hear the other side of the call, not just you).

You sign in with a magic link, Google, or Microsoft. The desktop app opens your browser to sign in, and the session deep-links back into the app when you finish.

1. Click Sign in in the app.
2. Your browser opens to intelliprompter.com/sign-in.
3. Enter your email and click Continue with email, or pick Google or Microsoft.
4. Click the link in your email (or finish the OAuth flow).
5. You land at /auth/app-bounce, which redirects back into the desktop app and signs you in.

Before you can use real-time coaching or Ask AI, configure at least one provider in Settings → AI Providers. You have four options:

• Anthropic Claude. Best coaching quality. Pay-as-you-go.
• OpenAI GPT. Solid. Pay-as-you-go.
• Google Gemini. Generous free tier.
• Ollama. Fully local, free. Requires Ollama running on your machine.

To add a provider:

1. Open Settings → AI Providers.
2. Click Configure on the provider you want.
3. Paste your API key (for cloud providers) or confirm Ollama is running.
4. Save. The provider shows Connected.

You can configure multiple providers. Smart routing automatically picks the cheapest capable model for each task: classification calls go to Llama for free if Ollama is configured, response generation goes to Sonnet, the hard reasoning passes go to Opus or GPT-4.

1. Click New meeting from the dashboard.
2. Give it a title.
3. Pick a Meeting profile. Profiles control which coaching modes are active and how the AI sounds. If you do not have one, create a default in Profiles.
4. Pick a Privacy mode. Default is Full. See Privacy Modes for what each one does.
5. Optional: attach a script. Click Write a script, paste your prepared content, click Save script.
6. Click Start.

The meeting overlay opens. It is invisible to screen sharing, floats above everything (including fullscreen Zoom or Meet), and shows your transcript, AI coaching, and script side by side.

Talk for a few minutes. You should see:

• Live transcript on the left, with speaker labels (renameable).
• AI coaching cards on the right, appearing every ~20 seconds when there is something worth saying.
• If you attached a script, it scrolls in sync with your speech and highlights the line you are on.

To end the meeting, click End Meeting in the overlay.

You land on the post-meeting Review page. Everything is here:

• Full transcript, segment by segment, with speaker colors. Segments are editable.
• Coaching events that fired during the meeting. Each has feedback buttons (was this useful?) that train future suggestions.
• Snapshot bookmarks (any moment you flagged with Cmd+Shift+S during the meeting).
• Highlight tags you applied per segment, with a filter strip across the top.
• Summary and action items, generated automatically.

If you ended the meeting by mistake, you can rejoin from the review page.

• Privacy Modes. Three modes, one per meeting.
• More docs landing soon: AI Providers (BYOK setup), Coaching Modes (the eight modes), Meeting Profiles, Knowledge Base (per-meeting RAG).