Features

Live context in. Clear guidance out.

A Windows desktop assistant that turns live context into clear, concise guidance — visibly and under your control.

See pricing Download

AI Interview Assistant — Session

LIVE · 12:48Screen · ExcalidrawCall audioMicAI · ollama.comPauseStop

TranscriptAuto-scroll

CALL12:31

Let's do a design question. How would you build a URL shortener?

MIC12:32

Sure. What scale should I design for?

CALL12:32

Around ten thousand new links a second. Reads are maybe a hundred times that.

CALL12:33

…and it should keep working if one region goes down

Detected question

System designConfidence: high

Design a URL shortener for about 10k new links a second, with 100× more reads.

From call audio and screen · revised twice

RegenerateNot a questionPin

Screen text · Excalidraw

Requirements

- 10k writes / s

- read : write = 100 : 1

- keep links 5 years

- custom aliases (nice to have)

GuidanceConciseSpeakingStreaming

Numbers first

10k writes/s ≈ 864M links a day, about 1.6T over five years. A 7-character base62 code gives 3.5T, so there's plenty of headroom.

Components

Stateless API behind a load balancer
ID service hands out counter ranges per node, encoded as base62
Key-value store keyed by short code, replicated across regions
Cache in front for hot links, since reads are 100× writes

Trade-offs

Counter ranges never collide but need a coordinator; hashing needs a collision check

Context sent: question · screen text · last 2 min of transcript · your session notes

Capture

It hears and sees only what you choose.

Nothing is captured before you press Start. Pause and Stop are always available. No audio or image files are written to disk.

Microphone

Your voice, transcribed on-device with the built-in Windows speech engine.

Call audio

The sound your computer plays, captured with echo suppression so AI Interview Assistant does not hear itself.

Screen

Only the window or region you choose, read with on-device Windows OCR.

Guidance

Concise answers, shaped to the question.

Questions, detected

Ask a question directly, or let AI Interview Assistant detect one from speech or the screen.

Read or hear it

Answers are concise and can be read on screen or spoken aloud.

Prompt profiles

Profiles shape tone, length and format for coding, system design, behavioral and meeting questions.

Your context

Optional user context (role, job description, resume summary and more) is used only for the session and never logged.

AI providers

Your key, or included credits.

Bring your own key

You configure an OpenAI-compatible provider — for example Ollama Cloud — and supply your own API key. The key is stored only in Windows Credential Manager on your computer, and the desktop app sends those requests directly to your provider. AI Interview Assistant's cloud backend never sees the key or those requests, and no credits are used.

Included Credits

Plans that list included AI credits can use Included Credits mode instead. AI Interview Assistant uses approved external AI providers to process Included Credits requests: the desktop app sends the request to the AI Interview Assistant backend, which checks your plan and credits and forwards it to the provider; the answer streams back the same way.

AI Interview Assistant does not run its own AI models and does not store the content of these requests. Credits are charged from the usage the provider reports, failed requests are not charged, and a retried request is never charged twice.

Privacy by design

Your sessions stay yours.

Capture and AI states are always visible and disclosed.
History, transcripts and guidance stay on your computer unless you export them.
The cloud backend stores only account, subscription, usage and audit data — never interview content. Included Credits requests pass through it to the AI provider without being stored.

Ready for your next practice session?

Choose a term, sign in on your PC and press Start.

See pricing Authorized use