Maiku AI listens to your interviewer during any video call and delivers real-time AI suggestions — completely invisible on screen share.
Live Demo — try the persona toggle
Video Interview — Google Meet
Sarah Martinez
Sr. Engineering Director · Stripe
You
Live Transcript
AI Suggestions
Waiting for question…
Response Style
This is what Maiku AI looks like on your screen — your interviewer never sees it.
Under the Hood
Every engineering decision optimised for one thing: giving you an edge without anyone knowing you have help.
Uses WDA_EXCLUDEFROMCAPTURE — the same OS-level flag banks use for DRM — so Zoom, Teams, and Meet record a blank window, not your AI copilot.
Your screen
Interviewer's recording
Audio → Whisper → LLM → suggestion in under half a second.
< 500ms
SQLite + BM25 full-text search. Your resume, notes, and docs stay on your machine — never uploaded.
Swap providers at runtime — zero config. Start free with Groq, upgrade to GPT-4o or Claude when you need it.
48 kHz system audio — hears your interviewer through any headset, no microphone permission required.
Quiet while you're speaking. Fires instantly when a question lands. Debounce windows shrink from 8s to 3s for rapid-fire rounds.
Drag the slider to see exactly what the interviewer's screen share recording captures — spoiler: the overlay vanishes entirely thanks to WDA_EXCLUDEFROMCAPTURE.
Mention O(n log n) merge sort here
How It Works
Four steps from download to invisible AI copilot in your next interview.
Step 01 — Create Account & Download
Maiku-AI-Setup.exe
125 MB · Windows x64
Backend started · Listening on port 8765
Type a real interview question or situation and watch Maiku respond in real time — powered by LLaMA 3.3 70B via Groq.
AI response will appear here in real time…
A random interview question fires — you have 60 seconds to answer before Maiku reveals the optimal hint.
The question will be read aloud. You have 60 seconds. When time is up, Maiku reveals the optimal approach.
In the real desktop app, hints appear automatically while you speak — no need to ask.
Testimonials
Join thousands of engineers and PMs who landed their dream role.
“I had Maiku running during my system design round. When the interviewer asked about database trade-offs I saw the exact talking points from my notes appear. I got the offer.”
Arjun Sharma
Senior SWE — landed at Stripe
“The invisible overlay is genuinely magic. I was on Google Meet and could see suggestions in real time without the interviewer having any idea. Zero stress, full focus.”
Emily Chen
Product Manager — hired at Linear
“Used Groq's free tier for 3 interviews. The latency is insane — suggestions pop up mid-question. Converted to Pro after I got the job. Worth every cent.”
Marcus Williams
Frontend Engineer — joined Vercel
“I uploaded my entire portfolio and past project notes. When the interviewer asked about my ML work, Maiku surfaced the right stats from my resume instantly.”
Priya Nair
Data Scientist — offer from Anthropic
Move the sliders to see your personalised return on a $24/month investment.
Industry studies show AI-assisted preparation improves technical interview pass rates by 30–60%. Use 40% as a conservative baseline.
Expected value gained
$3.0k
ROI multiple (3 months)
42x
Break-even in
1 days
vs. $24 / month cost
$24
Maiku pays for itself 42× over 3 months. Break even in 1 days.
Pricing
Start free. No credit card required.
All core AI features. Bring your own Groq API key. Open source on GitHub.
Unlimited interviews, every AI provider, guided setup wizard.
Everything in Pro, forever. Early adopter pricing — price rises as we grow.
All plans include a 14-day money-back guarantee. See full feature comparison
FAQ
Yes. Maiku uses the Windows Display Affinity API (WDA_EXCLUDEFROMCAPTURE) — a Windows-level system call that marks the window as excluded from any capture. This is the same mechanism used by DRM-protected content in browsers. It is not an overlay trick; the window is physically absent from the captured frame.
Yes. Maiku captures system audio via WASAPI loopback, which works at the OS level independently of which app is playing the audio. It works with any video call platform including Zoom, Google Meet, Microsoft Teams, Webex, and Discord.
A Windows 10/11 PC with x64 architecture. The installer bundles everything — no Python, no Node.js, no drivers required. Just download, install, and add your API key.
Groq (free tier — LLaMA 3.3 70B), OpenAI (GPT-4o, GPT-4 Turbo), Anthropic (Claude 3.5 Sonnet), and OpenRouter (100+ models). You bring your own API keys — they are stored locally on your machine and never sent to our servers.
Yes. The free plan uses Groq's free API tier which offers 7,200 seconds of Whisper transcription per day — more than enough for a full interview. You only need a free Groq account at console.groq.com.
When you upload a document (resume, notes, cover letter), Maiku indexes it locally with SQLite + BM25 keyword retrieval. During the interview, it automatically surfaces the most relevant snippets from your documents to include in AI suggestions — no internet needed for retrieval.
Windows may flag Maiku AI because it is a new publisher without a paid certificate yet. The full source code is public on GitHub — it is completely safe. Three easy fixes: (1) SmartScreen "Windows protected your PC" — click More info then Run anyway. (2) Smart App Control blocked (Windows 11) — open Windows Security → App & browser control → Smart App Control settings → set to Off, then re-run the installer. Windows Defender stays fully active. (3) Right-click the downloaded .exe → Properties → check Unblock at the bottom → OK, then run.
No. Audio is only sent to the STT provider you choose (Groq/OpenAI Whisper). The transcript and suggestions are processed between your machine and your chosen LLM API. Maiku's servers only handle user account data (email, plan). Your interview audio never touches our infrastructure.
The overlay does not generate any audio or interact with the microphone input — it only displays text visually on your local screen. It has no effect on the interviewer's environment whatsoever.
Contact
Bug report, feature request, or partnership inquiry — we reply within 24 hours.