Acexcode
A sub-second interview copilot that runs invisibly during the call.
A stealth real-time copilot that helps candidates walk into technical interviews armed with sub-second AI answers.
Build an interview copilot that lives on the candidate’s machine, listens to the call audio, reads the shared screen, and replies with structured answers in under a second — while staying invisible to the interviewer’s screen share.
- Dual-stream audio pipeline — microphone + system audio with under 300ms capture-to-transcription latency, powered by Whisper running locally for the fast path and a hosted instance for accuracy.
- Vision + reasoning loop — screenshots of the shared screen are passed through Gemini 2.0 Flash Vision and re-grounded against the candidate’s resume + the job description.
- OS-level stealth overlay — a click-through transparent surface that doesn’t appear in screen captures on Windows 10+. Toggling visibility takes under 50ms via a global hotkey.
- Multi-language code panel — syntax highlighting for JavaScript, TypeScript, Python, Java, Go, Rust, C++, and C#. One keystroke copies the suggested solution.
- STAR-format behavioural mode — the model is steered into Situation / Task / Action / Result structure whenever it detects a behavioural prompt.
Candidates press Ctrl + Shift + Space to start a session. The overlay listens silently, detects the question type (technical, behavioural, coding), and renders an answer the candidate can read off the screen — or paraphrase. Ctrl + B hides everything in the time it takes to glance at the interviewer’s webcam.
Trusted by candidates interviewing at every name on the FAANG roster — Google, Amazon, Meta, Microsoft, Apple, Netflix — plus high-bar coding platforms like HackerRank, CoderPad, and Codility. Live on Windows; macOS in private beta.