Built from your CV, projects & STAR stories · Local-first
interviewco.ai helps your answers
sound like you.
Drop in your CV, projects, and STAR stories. Mock interviews dig into your real background — so the answers sound like you, not a chatbot.
Profile grows with every mock·Local-first by default·Only the relevant snippet is sent to the model
REC · 04:51
‹
Q3 of 3 · currentaudio
Design a rate limiter for an API gateway handling 5k RPS.
I'd use a token bucket per API key in Redis for shared state across gateway instances. Token bucket gives smooth bursts; fixed-window has thundering-herd at boundaries.
Two-tier design:
- Local in-process counter — absorbs the bulk, ~50ns lookup.
- Redis Lua INCR+EXPIRE — atomic authority, syncs every 100ms
Token bucketRedis LuaTwo-tierBackoff hint
9:41•••
MacBook-Pro0.8s
ANSWERS · 3tap to expand
AUDIO● LIVE▲
Design a rate limiter for an API gateway handling 5k RPS.
Token bucket per API key in Redis. Two-tier: local counter, Redis Lua INCR+EXPIRE syncs every 100ms…
AUDIO32s ago▼
Tell me about yourself.
↓ Stream · in sync
0.8s
See it in action
One session. Two devices. Zero detection.
Watch the round-trip — question heard on your Mac, answer streaming to your phone.
ZoomMeetingEditViewWindowWed 9:41 AM100%
Zoom · Loop · Backend Engineer
Sarah K.· EM · TechCo
You· Candidate
“Design a rate limiter for an API gateway handling 5k RPS.”
MuteVideoShareChatLeave
REC · 04:51Q3 · currentListening
AUDIO IN
transcribing…
Design a rate limiter for an API gateway handling 5k RPS.
Stripe · rate-limit infraSystem design · L6Redis at scaleBackend resume bullet
⌥
■
♪
⊞
⟳
↑
9:41•••
MacBook-Pro0.8s
ANSWERS · 2tap to expand
AUDIOpending…
Design a rate limiter for an API gateway…
AUDIO32s ago
Tell me about yourself.
What the interviewer sees
You. Thinking. That's it.
Why Interview Companion
Four things no other real-time assistant gives you.
Real-time is table stakes — everyone has it. The real differences are architectural: where the data lives, how the profile evolves, where the answer shows up.
01
Local-first data
Audio buffers, screen pixels, your full profile, every transcript — all live on your Mac. There is no Interview Companion cloud bucket of your interviews.
02
Profile that learns
Every interview adds new fragments and signals which ones win. Your copilot gets sharper each loop — a knowledge base, not a one-shot prompt.
03
Answers shaped by you
Relevant resume bullets, projects, and behavioral stories are auto-selected per question. Answers sound like you, not like a generic LLM.
04
Mac + iPhone, in sync
Both apps sign in with your account. Mac listens and generates; your iPhone subscribes to the same session and shows answers as they stream — wherever you are.
macOS Core Audio Tap
nativeNative Sonoma+ system audio capture. No BlackHole, no virtual-driver hacks. Catches both sides of the call.
Screenshot solve
visionOne shortcut screenshots your coding window. Vision model returns a working solution with comments and edge cases.
Always-fresh models
managedWe route requests through our backend to the best-fit frontier model. No keys to manage, no model selection to second-guess — always current.
How it works
Mac captures. Profile contextualizes. Phone delivers.
STEP 01
Your Mac listens
System audio tap and optional screen capture pick up the question — spoken, typed into shared chat, or visible on screen. Transcription stays local until you say go.
On-deviceSTEP 02
Your profile contextualizes
Relevant fragments — resume bullets, behavioral stories, project notes — are auto-selected for the question and sent as context. The model answers as you, not as itself.
On-deviceSTEP 03
Your phone shows the answer
Streams to your iPhone or browser via your signed-in session in 2–3 seconds. Read naturally, in your own voice. The interviewer sees you thinking, not a teleprompter.
Covert readUse cases
Built for the full interview loop.
Not just LeetCode. Every stage of a senior engineering loop has a different shape — Companion adapts.
01
Coding rounds
LeetCode-style. Screenshot the problem; get a working solution with complexity notes and edge cases.
02
System design
Whiteboard rounds. Structured outline plus tradeoffs in your domain — sharded, replicated, observable.
03
Behavioral
STAR-formatted from YOUR stories — not generic AI-flavored anecdotes about “challenging stakeholders.”
04
Recruiter screen
Comp targets, locations, motivations one screen away. Surface them mid-call without scrambling.
05
Take-home review
Talk through your submission with confidence — even months after writing it. Companion remembers what you don't.
Privacy & control
You decide what leaves your Mac.
We're a local-first app, not a SaaS that hoovers your interviews. Here's exactly what happens — and what doesn't.
✓
Stays on your Mac
●Raw microphone & system audio buffers
●Screenshot pixels (only the question crop is sent)
●Resume, profile fragments, behavioral stories
●Full session transcript & answer history
●Account session tokens for your iPhone / browser
↗
Sent to AI providers
●
The transcribed question text
Forwarded through our backend to the model provider
●
The cropped question screenshot (vision mode only)
●
Relevant profile fragments selected for this question
●
Model parameters (temperature, length)
TLS end-to-end. Our inference backend forwards the request to the model provider — we don't retain content beyond the request, and we always send the provider's no-training flag.
✕
Never. Not now, not ever.
Never used for training
Your transcripts, answers, and profile are not used to train any model — ours, our AI providers', or anyone else's. We send API requests with the standard no-training flag.
Never sold or shared
No third-party analytics on session content. No data brokers. No “anonymized aggregate” loopholes.
Never recorded by us
No audio or video storage server on our side. Question text passes through our inference proxy but is not retained beyond the request — there is no archive of your interviews on our infrastructure.
Never read by humans
Engineering doesn't have access to your in-flight content. Logs strip session payloads, and support requests don't include transcripts unless you attach them manually.
Delete anytime
Wipe your profile or history in one click.
Your whole profile lives on your own machine — nothing is stored on our servers. Delete it anytime: clear it from the Privacy panel in one click, or just remove it yourself.
Delete session history
Removes all past transcripts and answers. Profile stays.
Reset profile fragments
Clears resume, behavioral stories, projects. History stays.
Sign out of synced devices
Revokes the session on all phones & browsers signed in to your account.
Delete everything
Full local wipe. Equivalent to a fresh install.
No cloud by defaultTLS to providersNo training on your dataLogs strip session payloadsGDPR / CCPA aligned
Why not them
The market is loud. We're different where it matters.
Price is the easy battle. The real differences are architectural.
| Interview Companion | Final Round | LockedIn | Interview Coder | ChatGPT | |
|---|---|---|---|---|---|
| Mac local-first capture | ✓Native Core Audio Tap | —Cloud upload | ~Web app capture | ~Screenshot only | —Manual paste |
| Private phone display | ✓Web receiver + Mac overlay | —— | —Web tab only | —Mac overlay | —— |
| Screenshot + audio together | ✓Same session | —Audio only | ~Separate flows | ~Screenshot only | ~One at a time |
| Structured candidate profile | ✓Per-question fragments | ~Resume paste | ~Resume paste | —— | —Per-chat upload |
| Technical interview focus | ✓Coding · sys design · BQ | ~Behavioral-heavy | ~General | ✓Coding only | —General-purpose |
| Answer summary language | ✓Pick your summary language | —English only | —English only | —English only | ~Prompt workaround |
| No browser extension needed | ✓Standalone Mac app | —Chrome extension | —Chrome extension | ✓Standalone | —Chrome extension |
| Privacy model | ✓Local-first, no cloud bucket | —Cloud storage | —Cloud storage | ~Cloud screenshots | ~Cloud + opt-out |
| Price | $49–69 / mo | $150 / mo | $120 / mo | $299 / mo | $20 / mo |
Based on each product's public marketing as of May 2026. We'll gladly correct anything misrepresented — file an issue on our GitHub.
Pricing
One plan. Two ways to pay.
Private by default. Try free, upgrade when you have a real interview — cancel anytime.
Free
$0forever
Practice with mock interviews. No real-interview access.
- ✓3 mock interviews (lifetime)
- ✓General model
- ✓30-min session cap
- ✓1 mock per day
- —Real interview mode
- —iPhone & web delivery
Pro Monthly
$69per month
≈43% cheaper than LockedIn
Full copilot. Flexible month-to-month.
- ✓Real interviews — up to 60 / mo
- ✓Premium AI models + screenshot vision
- ✓iPhone, web & Telegram delivery
- ✓90-min session cap, 5 / day
- ✓Unlimited mock interviews
- ✓V1 batch transcribe pipeline
BEST VALUE
Pro Quarterly
$49per month, billed $147 / quarter
≈39% cheaper than LockedIn quarterly
Same features. Save 29% with a quarter commitment.
- ✓Everything in Pro Monthly
- ✓Up to 50 sessions / mo
- ✓150 sessions / quarter cap
- ✓Best fit for active job-search cycle
- ✓Lower per-session cost
- ✓Priority support
All sales final. 14-day right-of-withdrawal for EU customers under Distance Selling Directive.
1 session = up to 90 min of real-interview audio + unlimited screenshots within that session. Mock interviews don't count.
FAQ
Honest answers to the real questions.
Companion runs entirely on your Mac — no browser extension, no virtual camera, no proctoring-hook integration. There is nothing on the candidate machine for Zoom, HackerRank, CodeSignal, or HireVue to detect beyond a generic native app. That said, aggressive eye movement, glancing patterns, or pasting answers verbatim are detectable by humans regardless of any tool. Use it as a copilot, not a teleprompter.