Ask this next · live interview copilot · macOS

Every bad hire sounded good in the interview.

Caliber gives you the follow-up question that reveals whether the candidate actually did the work — live, while they're still in the call.

Built for founders, hiring managers, and recruiters running technical interviews.

2 interviews free, up to 7 days · macOS 12+
Download for Windows Beta
Early access · you'll see a Windows “Run anyway” prompt (signed build coming soon) · Windows 10/11
caliber · live sessionv0.5.24 Live
Live transcript · Ask this next
Interviewer: Walk me through how you'd shard a write-heavy table.
Candidate: You'd typically choose a shard key with high cardinality and even access distribution, then route writes through a consistent-hashing layer…
✦ Ask this next · stretch
"Re-derive the shard key if writes are 10× read-heavy and one tenant is 80% of traffic — what changes?"
Reason: reads rehearsed — over-polished, no hesitation.
Rubric coverage
System design
Communication
Ownership
Problem solving
7.4
Overall · scorecard ready
Signed + notarized by Apple
Invisible during screen share
60-second scorecards
Live next-question probes
Signature capability

The best next question — while you can still ask it.

Anyone can review an interview afterward. Caliber works during it: as the candidate answers, it drafts the single sharpest follow-up to ask next — the one that forces original recall instead of a rehearsed or memorized line.

  • Verify · ownership · constraint. Every probe does one job: test whether the reasoning actually holds up, separate their work from the team's, or change a constraint and make them re-derive it live — the hardest thing to fake.
  • Ready before they finish. The question builds as the candidate talks, so you have one in hand mid-answer — plus a fresh probe on the whole answer the moment they stop.
  • It tells you which answer to probe. Caliber quietly flags the answer that suddenly reads like an LLM — so the follow-up lands exactly where it matters.
  • A suggestion, not a script. Ask it as-is, reword it, or ignore it. Caliber proposes; the judgment stays human.
Which answer to probe live
Answer 1 · system designoriginal
Answer 2 · debuggingoriginal
Answer 3 · the migrationProbe recommended
Answer 4 · trade-offsoriginal

Answer 3 reads rehearsed — and Caliber already has the probe that breaks the act.

Why interviewers miss signal

The problem isn't the first answer. It's the follow-up.

Candidates prepare stories. AI helps polish them. The first answer sounds great — and tells you almost nothing about whether they actually did the work.

Caliber gives you the follow-up that reveals:

  • Ownership. Was it their decision, or the team's?
  • Trade-offs. What did they give up — and why?
  • Reasoning. Does the logic still hold when you change a constraint?
  • Real experience. What actually broke — and how did they know?
"Anyone can rehearse the answer. Nobody rehearses the follow-up."
That's the gap Caliber closes — live, while you can still ask it.
See it work

The follow-up that changes the answer.

A rehearsed answer survives the first question. It rarely survives the right second one.

Live interview · networking migration
Candidate · "I migrated our entire networking layer to gRPC last year."
✦ Ask this next
"What was the hardest production issue you hit during that migration — and how did you catch it?"
The story was rehearsed. The follow-up isn't. This is the moment you find out whether they actually shipped it — or just read about it.
Real probes

Three real probes from production.

The kind of follow-up Caliber surfaces live — mid-answer, before the candidate can pivot back to the script.

Candidate · "I scaled our image pipeline."
✦ Caliber asks
"What broke first when memory pressure increased?"
Candidate · "I led the migration."
✦ Caliber asks
"Which decision did you personally make that the team disagreed with?"
Candidate · "We reduced crashes by 40%."
✦ Caliber asks
"Which crash remained unsolved the longest?"
How it works

From CV to scorecard, in five steps.

Set up in under a minute, then let Caliber run quietly beside you for the whole interview.

  1. 1

    Add the candidate

    Drop their CV (PDF, DOCX or image) and Caliber reads the name, role and key claims and suggests a rubric — or type it in, or paste a LinkedIn URL. You get a pre-interview brief before the call.

  2. 2

    Set your rubric

    Confirm the dimensions you'll score on — use the suggested set or your own. This is what every answer gets measured against.

  3. 3

    Record & run the interview

    Hit record and talk. Caliber transcribes both sides live, knows who said what, and stays invisible during screen share — Zoom, Meet and Teams never see it.

  4. 4

    Caliber hands you the next question

    As the candidate talks, Caliber drafts the sharpest follow-up to ask next — ready before they finish, plus a fresh one on the whole answer when they stop. Rubric coverage fills in, a quiet flag tells you which answer to probe, and you can still ask Caliber anything by voice or text.

  5. 5

    Generate & export the scorecard

    One click turns the whole conversation into a structured scorecard — verdict, dimension scores, strengths and follow-ups — in about 60 seconds. Export the PDF to your ATS and draft follow-up emails.

Capabilities

A second interviewer focused on what matters.

Seven capabilities running in one invisible window — from the brief you read beforehand to the PDF you drop into your ATS.

Live transcription

Both sides of the call, diarized in real time. Caliber knows who said what — interviewer vs. candidate — with sub-second latency.

Ask this next

Caliber drafts the sharpest follow-up as the candidate talks — verify a claim, test ownership, or change a constraint and make them re-derive it live. A ready question in hand, mid-answer.

AI-answer detection

Under the hood, Caliber compares each answer to the candidate's own earlier ones. When a later answer suddenly reads like an LLM, it flags it — so you know exactly which one to probe.

Live rubric coverage

A heads-up display that fills in as you probe each dimension — and a next-question nudge so you never leave the call with gaps.

60-second scorecards

Click "generate" and Caliber turns the entire conversation into a verdict, dimension scores, strengths, follow-ups and a PDF — straight into your ATS.

Vision mode

Reading a CoderPad or a shared design? Caliber sees the screen too, so its read of the answer includes the code on the candidate's screen.

Invisible during share

Excluded from every macOS screen-capture API. Zoom, Meet, Teams, QuickTime — none of them see Caliber. Only you do.

Trust & privacy

Built for interviews that involve real PII.

You're handling resumes, names and live conversations. Caliber is engineered so that stays yours.

Encrypted sessions

Your sign-in is stored in the macOS Keychain. Scorecards sync over TLS, scoped to your account only.

You own deletion

One click wipes your account and every scorecard from the cloud. No retention games, no dark patterns.

Invisible by design

Excluded from macOS capture APIs at the OS level — not a hack, a hardened-runtime entitlement.

Enterprise-grade controls

SSO/SAML, data-residency control and SLA-backed uptime available on the Enterprise tier.

Why Caliber gets sharper

It measures what no one else does.

Most interview tools score what already happened. Caliber tracks which of its probes you actually use — the first real signal of how interviewers dig for depth.

  • Asked. You used the probe as-is.
  • Edited. You reworded it, then asked.
  • Followed. You asked the same idea in your own words.

Every interview turns into signal about how you interview — the foundation we're building personalization on. It's the moat a generic note-taker can't copy.

Probe outcomes · measured live
✦ "What broke at 10× traffic?"Asked
✦ "Whose decision was the rewrite?"Edited
✦ "Walk me through the rollback."Followed
Every probe is measured — learning from real interviews, not synthetic benchmarks.
Pricing

Pick the tier that matches how you hire.

From individual recruiters to hiring orgs running 100+ interviews a month. Free trial: 2 interviews, up to 7 days — no card required.

One bad engineering hire costs months — and tens of thousands.
Caliber costs less than a single hour of interviewing. Most teams make the subscription back on the very first mis-hire they avoid.

Solo

For individual recruiters

₹2,999 / month
For hiring managers running about 25 interviews a month.
  • All core features
  • 25 interviews / month
  • Live probes — "Ask this next"
  • Live AI-answer detection
  • Pre-interview brief + scorecard PDF
  • Email support · 48h response
  • Team features

Team Coming v0.6.0

For hiring teams

₹19,999 / month
5 seats included
For hiring teams running parallel interviews. Currently in build — join the waitlist for founders' pricing.
  • Everything in Pro
  • 5 user seats · 500 interviews / month pooled
  • Shared scorecard library
  • Team analytics dashboard
  • Admin panel + user provisioning
  • Onboarding session included
  • Priority chat support · 4h response

Enterprise

For hiring orgs

Custom
From $99/seat · 10-seat min
For organizations running 100+ interviews a month with compliance needs.
  • Everything in Team
  • 10+ user seats
  • SSO / SAML
  • SLA-backed uptime
  • Custom rubric library
  • Dedicated CSM
  • Data residency control
  • Custom contracts + invoicing
Contact sales

Compare features across tiers

Solo Pro Team Enterprise
Core product
Live transcription
Live probes — "Ask this next"
Pre-interview brief (resume + LinkedIn)
Live rubric coverage HUD
End-of-call scorecard + PDF export
Live AI-answer detection
Vision mode (screen reading)
Invisible during screen share
Capacity
Interviews / month25500 pooledCustom volume
User seats1510+
Team & collaboration
Shared scorecard library
Team analytics dashboard
Admin panel + user provisioning
Enterprise readiness
SSO / SAML
SLA-backed uptime
Custom rubric library
Data residency control
Dedicated CSM
Support
Email support48h4h1h SLA
Onboarding session✓ + workshop
Dedicated Slack channel

Need more capacity? Upgrade to the next tier — or talk to us about a custom plan. Caps are generous on purpose; we'll never throttle you mid-interview.

Pricing questions

How does the free trial work?

Download Caliber, install, and start your trial automatically. You get 2 full interviews or 7 days from install, whichever comes first. All features unlocked. No credit card. When you hit either limit, the app prompts you to subscribe — your data and settings stay intact.

Can I switch tiers later?

Yes — upgrade or downgrade anytime from the app. Upgrades take effect immediately and we prorate the difference. Downgrades take effect at the end of your current billing period.

What counts as an "interview"?

An interview is a single session: you click the mic, run the call, then click "Generate scorecard" at the end. Drafting a brief, replaying a transcript, or exporting a PDF doesn't count — only the live interview session itself. The counter resets at the start of each monthly billing cycle.

What happens if I hit my monthly cap?

You'll see a heads-up at 70% of your cap and again at 90% so it never sneaks up on you. If you hit 100%, Caliber asks you to upgrade — but we'll never throttle you mid-interview. The session you're already running always finishes and exports a scorecard. Solo gets 25 / month, Pro gets 150 / month, Team gets 500 / month pooled across your 5 seats, and Enterprise is sized to your volume.

Paid charges aren't refunded after upgrade — the 7-day trial is your evaluation window — but you can cancel anytime from inside the app and the next charge stops immediately. Full text in the refund policy.

2 interviews free, up to 7 days

Your next interview, read in real time.

Install it, run one call, and watch the scorecard write itself. No card. No setup. Just you, sharper.

Download Caliber for Mac
macOS 12+ · Apple Silicon & Intel · Signed + notarized