Private desktop copilot

Invisible AI coaching for interviews, tuned to your voice now

Native audio capture. Ghost overlay only you see. Free beta now, then a one-time purchase. Works in Zoom, Teams, Slack, and any desktop app.

Get the Beta for Windows See How It Works
Available for
macOS & Windows
Listening…
Why Diwa Wins

Built different. Built for control.

A native desktop copilot engineered for privacy, speed, and zero setup.

Native Audio, No Virtual Cables

Works where browser extensions fail. Zoom, Teams, Slack, Discord—any desktop app. Native OS audio capture with zero setup.

macOS Native
Windows WASAPI
No drivers needed

Zero Telemetry. Your Data Stays Yours.

Bring your own key. No account. No cloud uploads. No analytics. Your API key is encrypted locally and sent directly to OpenAI.

0 Servers
0 Telemetry
100% Local
BYOK Architecture

One-Time Purchase, Not $99/month

Free during Beta, then a one-time purchase. You pay OpenAI directly—usually pennies per interview.

$99/mo Others
→
Free* Diwa Beta

11+ AI Models. Your Choice.

Switch between GPT-4o, o3, and Realtime on the fly. Choose speed or depth based on the moment.

GPT-4o o3 Realtime +8

Ghost Mode — Truly Invisible

Click-through transparency. Invisible in screen shares. Only you can see it.

Zoom Meeting
Diwa Overlay Invisible to others

Diwa vs. Browser Extensions

Compare a native desktop copilot with the usual browser-based approach.

Diwa Copilot
Browser Extensions
Platform Support ✓ Any desktop app
Zoom, Teams, Slack, Discord 		✗ Chrome/Edge only
Won't work in native desktop apps
Audio Capture ✓ Native OS loopback
No virtual cables 		✗ Virtual cables or tab audio
BlackHole / VB-Cable setup required
Data Privacy ✓ 100% local storage
No cloud, no telemetry, BYOK 		✗ Cloud servers
Data processed on vendor infrastructure
Cost ✓ One-time purchase
Free Beta + pay only API usage 		✗ $50–$200/month
Subscription + usage limits
Stealth / Ghost Mode ✓ OS-level click-through
Invisible to all screen sharing 		~ Browser overlay
May be visible in some screen shares
AI Model Selection ✓ 11+ models, user-selectable
GPT-4o, o3, Realtime, Codex & more 		✗ Fixed model, vendor-chosen
No user control over AI backbone
Latency Control ✓ Dual pipeline (~200ms or batched)
User switches between speed & quality 		✗ Single pipeline
No user control over latency
Custom Personas ✓ Unlimited custom personas
Role + company + personality + style 		~ Single profile
Limited or no personalization
Get the Beta — No Subscription

No credit card. No sign-up. Just your OpenAI API key.

4 Simple Steps to Get Started

No account needed. Just your OpenAI API key.

Step 01: Download & Install
Step 02: Add Your API Key
Step 03: Upload Context
Step 04: Start Interview

Common Questions

Transparent answers for a transparent tool.

🚨 "Unrecognized app" warning?

We're in early Beta without code-signing certificates yet.
macOS: Right-click → Open.
Windows: Click "More Info" → Run Anyway.

🔑 Is my API key safe?

Yes. Your key stays encrypted on your device via OS Keychain and goes directly to OpenAI. We have no servers—nothing to leak.

🎧 Why Screen Recording on Mac?

macOS needs this permission for native system audio capture. No BlackHole or virtual cables required. Relaunch Diwa after enabling.

🪟 Audio on Windows?

Native WASAPI loopback—just click "Start Capture." No screen picker, no setup.

💰 Really free?

Free during Beta (until May 1, 2026). You pay only OpenAI API usage—usually pennies per interview.

Stop blanking. Start closing.

Get the free Beta. Bring your OpenAI API key. Own the interview.

For your safety, download only from this website. Third-party sources may be unsafe.

v1.7.0 · Requires OpenAI API key · 50 downloads