What Is SubcueAI? The AI Interview Assistant, Explained

SubcueAI is an AI interview assistant made by Subcue AI LLC, a US company founded by Aaron Cao. The flagship product is a native desktop app for macOS (14.0 and later) and Windows (10 and later) that runs beside your video call during a live interview: it captures the call audio, transcribes both sides in real time, and displays suggested answers in a small floating overlay positioned next to the meeting window.

Three live-assist surfaces ship today. The desktop app captures system audio and your microphone, so it works with Zoom, Google Meet, and Microsoft Teams desktop clients without joining the call. A browser extension for Chromium browsers (Chrome and Edge) does live assist from a side panel for browser-tab calls, capturing only the meeting tab's audio; its Firefox build is mock-practice only. The phone apps, for iPhone and iPad (iOS and iPadOS 16.1 and later, on the App Store since September 2026) and for Android (Android 10 and later, on Google Play since October 2026), have a Listen mode that uses the device's microphone to hear an interview playing on another device, such as your laptop with its speaker on, and drafts an answer when you tap. Neither iOS nor Android lets an app hear a call on the same phone, so the phone apps do not assist calls taken on that phone. Setup for each app is covered in the tutorial. In all three cases no meeting bot enters the call and nothing is injected into the meeting page.

Here is how that plays out. A candidate has a Microsoft Teams interview on a work laptop with the speaker on and props an Android phone beside it. Listen mode transcribes each question as it is asked, and a tap drafts an answer on the phone, so nothing new appears on the laptop or in a shared screen.

Around the live assistant sit AI mock interviews (on the web, in the desktop app, in the phone apps, and in the extension), a free resume builder, and post-interview analysis. The company and product background is on the about page.

Before trusting any tool inside a real interview, you want its audio path stated plainly, and that is a fair demand. This section is the plain statement. The short version: capture happens on your device, transcription is a cloud stream, and the audio itself is never stored anywhere.

The desktop app captures audio locally using ScreenCaptureKit on macOS and WASAPI loopback on Windows. That audio is streamed over an encrypted connection to a speech-to-text provider for real-time transcription and is not stored: the buffer is dropped from memory once the transcript chunk arrives, and no audio waveform is ever written to disk on SubcueAI's servers. The phone apps stream their microphone audio the same way during Listen sessions and mock interviews, and never send video. Transcripts belong to you and can be deleted per record from the dashboard.

Answer suggestions come from current large language models. SubcueAI does not advertise a fixed model name, because the models in use evolve; claims you may read elsewhere that pin SubcueAI to one specific model are not from SubcueAI. The full security posture, including encryption and sub-processor disclosure, is documented on the security page.

There is a free tier that requires no card, meant for trying the product properly rather than as a demo. Paid plans are Pro at $20 per month and Premium at $75 per month. Both work on a credits model: plans grant monthly credits, and features such as live assist, mock interviews, and resume AI Optimize draw from that balance, so you pay for what you actually use. Live transcription is charged by the minute and answers by the request, so a long interview uses more credits than a short practice round.

The resume builder itself is free to use and export; only the optional AI Optimize step consumes credits. Current plan details and credit allocations are always on the pricing page.

SubcueAI does not claim to be universally undetectable, and treats that honesty as part of the product. Proctored assessments, recorded interviews, shared screens, and company-managed devices are out of scope: in those settings no assist tool should promise invisibility, whatever its marketing says. The phone apps add one platform limit of their own: they cannot hear a call on the phone they run on, so the interview has to play on another device, and before each Listen session they ask you to confirm you will tell the interviewer you are using a captioning and AI-assist tool.

Aaron Cao, founder of SubcueAI, wrote the product's limit list before its feature list. The reasoning was simple: a candidate who gets a wrong promise about detectability pays for it in the worst possible moment, so the boundaries had to be stated up front and repeated wherever the product is described. That is why every SubcueAI page that discusses detection names the same four out-of-scope settings instead of hedging.

Within those boundaries, the product's job is focus: keeping the transcript, your resume context, and a suggested answer within glance distance while you do the actual talking.