ModelFoldModelFold
Get Started
Desktop app

Interview Assistant

Captions appear as soon as they speak; a suggested answer arrives as soon as they ask. A real-time copilot for online interviews, meetings and client calls.

macOS · Apple silicon
M1 / M2 / M3 / M4 · macOS 15 or later
Free download.dmg · 47 MB
macOS · Intel
In progress
Coming soon
Windows
In progress
Coming soon
Version 1.0Requires a ModelFold API key (answers are billed per use, about 1 credit each)Get a key →Install guide
Live transcript / Q&A
10:02:11Could you start with a quick intro to your latest project?
10:02:19What was the hardest technical problem there, and how did you solve it?
❓ What was the hardest technical problem, and how did you solve it?
💡 Peak-hour API latency. I traced it to slow database queries, added caching and an async queue, and brought p99 from about 2s to roughly 300ms, then added load tests and alerts so it would not regress.

Features

Live captions

Captures the audio playing on your Mac, so it works with Zoom, Teams, Meet, Lark or any browser tab, even with headphones. Mixed Chinese and English, text about 1 second after each sentence.

Question detection

Finds the questions actually being asked, fixes recognition typos, and stays quiet during small talk.

Suggested answers

ModelFold chat models draft a 2–4 sentence, conclusion-first answer in the language of the question, without making up numbers.

Audio stays on your Mac

Speech recognition runs locally and offline (Whisper); audio is never uploaded. Only the recognised text is sent to the chat model.

Floating caption window

An always-on-top, translucent window that sits over your meeting app; drag, resize, and it remembers where you left it.

Full record afterwards

Every session saves the transcript and all Q&A, opened automatically when you stop, ready for notes and review.

Get started in three steps

Read the step-by-step guide →
  1. 1

    Download & install

    Open the installer and drag the app into Applications. If macOS blocks the first launch, click Open Anyway under System Settings → Privacy & Security.

  2. 2

    Add your API key

    Create an API key in your ModelFold account, paste it into the app, and pick the answer model and language.

  3. 3

    Press Start

    macOS asks for Screen & System Audio Recording once, and the speech model (about 1.6 GB) downloads on first run.

Installing for the first time?

Screenshots for every step: download, opening an app macOS could not verify, allowing audio recording, adding your key and getting started.

→

FAQ

macOS says Apple could not verify the app+

Expected for this beta, which is not notarized yet, and only needed once: click Done in the dialog, open System Settings → Privacy & Security, scroll to the bottom, click Open Anyway and confirm with your password.

Why does it need Screen & System Audio Recording, and how do I allow it?+

macOS only lets apps record other apps’ audio through this permission; the app records system audio only, never your screen. The first time you click Start, follow the prompt to System Settings → Privacy & Security → Screen & System Audio Recording, turn the app on, then reopen it.

How much does it cost?+

The app is free. Suggested answers use chat models on your own ModelFold account and are billed by actual usage: about 1 credit (≈ $0.005) per answer with the default model, and only when a question is asked.

Do I need to install anything else?+

No. The runtime and speech engine are bundled; the speech model downloads once on first run.

Which computers are supported?+

Macs with Apple silicon (M1 or later) on macOS 15 or later. Intel Macs and Windows are in progress.

Is my audio uploaded?+

No. Speech recognition runs offline on your Mac; only the recognised text is sent to the chat model when a question is asked.

Questions or feedback: support@modelfold.online.