Interview assistant: questions and answers
In short: below are answers to common questions about AI interview assistants: how they work, is it legal, does the employer notice, what resources are needed and how much it costs.
Is there a neural network for interviews that answers questions?
Yes. Such programs listen to the interviewer's question, recognize it and show a ready answer on screen in real time. You do the talking — the neural network works as a teleprompter. Neural network for interviews →
What is an interview assistant?
It's a program that recognizes the interviewer's question in real time and shows an answer hint on screen. It doesn't speak for you — you read the gist and answer in your own words. Details: AI at interviews →
How does an AI assistant help pass an interview?
It relieves anxiety and helps recall the answer structure: definition, example from practice, nuances. It works faster than searching for an answer in a browser by hand. Step-by-step instructions →
Does the assistant answer in my voice?
No. It's a teleprompter: the hint is shown as text on your monitor, and you speak yourself. A voice answer would sound unnatural and be noticed immediately. A teleprompter for interviews →
Can the employer tell I'm using an assistant?
On a regular call — no, if the window is hidden by Invisible Mode. Under proctoring, third-party tools can be detected, and if the rules ban them it's a violation. Invisible Mode →
Can I use an assistant in a proctored interview?
As a rule, no. If the rules explicitly forbid third-party tools, the assistant must not be used regardless of whether it is technically detected. In that case use it only for preparation. All about proctoring →
Do I need knowledge, or will the assistant do everything for me?
You need knowledge. The assistant suggests facts and structure, but personal experience and understanding come only from the candidate. The tool amplifies a prepared specialist, it doesn't replace one.
Do I need internet for the assistant to work?
Speech recognition can run fully offline. Internet is usually only needed for the language model request that generates the answer. Offline or cloud →
Does my data go to the cloud?
In Mockingbird — not the audio or the résumé: speech, the PDF and the knowledge base are processed locally. Only the LLM request text goes out. Privacy analysis →
What's the latency from question to answer?
On a GPU with CUDA — about 2–5 seconds. On CPU it's longer, so an NVIDIA card is recommended for comfortable use. Choosing a model →
Do I need an NVIDIA GPU?
Not necessarily, but it's desirable. Without CUDA, speech recognition runs on CPU — slower. The app switches to CPU automatically when no GPU is present.
Which speech recognition model should I choose?
The default is faster-whisper large-v3-turbo — the optimum of accuracy and speed on GPU. On weak hardware pick smaller models. Why large-v3-turbo →
Can I use an assistant in a technical interview with coding?
Yes, it helps with theory and explanations while you write the code yourself. Show your thinking instead of copying a ready solution — that matters for the assessment. How to answer questions →
Is there an interview bot that helps answer?
Yes, but full-fledged solutions aren't chatbots with manual input — they're programs with automatic speech recognition: they hear the question themselves and show the answer in 2–5 seconds. How it works →
Does the neural network answer in an interview by voice?
No — and it shouldn't. Voice synthesis sounds unnatural and is immediately noticeable. The working mode is a text hint on your screen. Teleprompter →
How much does a neural network for interviews cost?
Mockingbird is free and distributed as open source. You can download it from the GitHub releases page. Download →
Which platforms does the assistant support?
Mockingbird supports Windows 10/11 and Linux (x86_64). Installation →
Mockingbird — an offline interview assistant: local speech recognition, a knowledge base built from your résumé, answers in 2–5 seconds. Installation takes minutes.