Precognia listens to your live interview and puts the right answer on your screen — built from your own studies, batches and deviations, with the controls and the reasoning still in it.
Free to start · No credit card · Private to your account
"Tell me about a deviation you investigated. What was the root cause?"
We had an out-of-spec dissolution result on a stability pull — one tablet out of six, so not obviously a real signal.
I ran the OOS investigation before assuming lab error: re-tested with a second analyst, checked the media prep records, and pulled the chamber logs. The chamber had drifted about two degrees for eleven days during a power event, which nobody had flagged because the alarm threshold was set wider than the protocol.
The batch was fine on re-test but I widened the scope to every study in that chamber. What I'd defend is not writing it off as analyst error. What I'd change is that the alarm threshold and the protocol limit should never have been two different numbers
Example shown: pharma / QA. Precognia reads your field and your level from your own resume and job description — no role setup, no templates.
Not a generic behavioural list. These are the questions this interview asks, and the ones Precognia is answering from your own material while you're being asked them.
"Walk me through a study you designed. What were your controls?"
"Tell me about a deviation you investigated. What was the root cause?"
"How do you handle a result you can't reproduce?"
"Describe a time you had to defend your data to someone senior."
"What would you do if you found something two days before a shipment?"
"How do you keep documentation audit-ready under time pressure?"
Two candidates can describe the same work and only one of them gets the offer. The difference is what the answer is built out of.
An answer that says what the experiment ruled out — and what it couldn't — reads as someone who has actually been challenged on data before.
Knowing when to stop the line, hold the batch or flag the result is the judgment these interviews are testing. Every answer includes what you decided and when.
Answers are built only from the material you upload, so nothing you say is a claim you can't put a batch record or a protocol behind.
Precognia works out your field and your level from your resume and the job description, then answers the way someone one or two rungs up would — which is what makes an interviewer sit forward.
Describing what you ran is the baseline; anyone can read a protocol back. The rung up is design and defensibility: why those controls, what would have invalidated the result, what you'd have escalated and to whom, and how the documentation would survive an audit. Precognia pulls that out of your own work.
It can't invent experience you don't have, and it doesn't try — a claim you can't defend in the follow-up costs you more than the answer won. Everything it says comes from what you gave it.
The free plan is real sessions on a real interview, not a locked demo.
No credit card. See it work in one real question.
During early access you get more than this — 5 sessions, 30 minutes each — while we finish billing.
For an active job search.
Paid plans aren't open yet. You'll start on Free, and this is what Pro will include when it opens.
For heavy interviewing and multiple tracks.
Paid plans aren't open yet. You'll start on Free, and this is what Power will include when it opens.
* "Unlimited sessions" is bounded by your monthly AI-hour allowance. Most people never come close to 10 hours — that's roughly 10 full-length interviews.
Paid plans are billed monthly and you can cancel any time. Because AI compute is consumed immediately as you use it, subscriptions are generally non-refundable. If you think there's been a billing error or a technical problem, contact us and we'll review your case.
Add your resume and the job description, and Precognia is ready before your call starts.
Or read the FAQ and privacy policy first.