What is sherlockai?

Sherlock AI is a desktop application built to record and analyze video call interviews as they happen. It is meant for situations where an interviewer wants to check whether the person on the call is really who they claim to be and whether they are following the rules of the interview. The app has two parts working together. An Electron and React frontend handles the screen, audio, and camera capture along with the dashboard you interact with. Behind that sits a Python service that does the heavier analysis work: detecting and tracking faces, watching where the person's eyes are looking, transcribing speech, and reading any text that appears on screen. Once all of that data is collected, a connected AI model is used to write a report summarizing what was observed, including checks for possible plagiarism in spoken answers. To run it, you need Node.js and Python installed on your computer, along with ffmpeg for converting audio recordings. Setup involves installing the app's JavaScript dependencies, creating a Python virtual environment and installing its requirements, and adding an API key for OpenRouter so the AI report step can work. The first time you run it, several AI models for face recognition, speech recognition, and text reading are downloaded automatically, which takes some time and disk space. All of the interview data, including video, audio snippets, and the generated reports, is stored locally in a SQLite database on the machine running the app. The project currently lists Linux as its primary supported system, with macOS and Windows also listed as supported operating systems. This is a niche, specialized tool aimed at people running structured interviews who want an automated record of what happened on a call, rather than a general purpose video calling or note taking app. Because it combines several separate AI services behind one interface, it is best suited to someone comfortable installing and configuring developer tools rather than a plug and play consumer product.