There's a quiet pattern I've watched repeat for years in every company I've helped hire into. The candidate with the strongest credentials, the most directly relevant experience, the most thoughtful work samples — they lose the offer. They lose it to someone who interviews better.
Not someone who works better. Someone who interviews better. There's a real gap between those two things, and that gap is the most expensive talent leak in the world.
The interview gap
I'm Reda. I built Talentee because I got tired of watching this happen.
Before founding Talentee, I spent years on both sides of recruiting desks — as a candidate trying to break into roles I knew I could do, and as a hiring manager evaluating people I knew would be either great or disappointing. The pattern was always the same: in a 45-minute conversation, the candidate's actual ability to do the work matters less than their ability to articulate that ability under pressure.
That's not a moral failure of hiring. It's a structural problem. Interviewers have to make a high-stakes decision with limited time and incomplete information. The person who can structure their answer cleanly, ground it in a specific example, and project confidence has a massive edge. The person who knows the work cold but can't translate it into a polished narrative loses. Every time.
And the worst part: most candidates never get told what they're doing wrong. Rejection emails are vague. Coaches are expensive. Friends give well-meaning but generic feedback. So candidates iterate slowly, sometimes never improving the things that actually cost them offers.
What I tried before building this
The first version of "fix this gap" was me running mock interviews for friends, then for friends-of-friends, then for people I'd never met. I'd record the calls, watch them back, and write up 2-3 specific things that needed work — the long pause before the question was finished, the four-minute answer that should have been ninety seconds, the answer that named a behavior without naming the impact.
Three things became obvious fast.
One, the feedback that actually moved people wasn't about content. It was about structure under time pressure. People who could give a clean STAR answer in two minutes outperformed people with stronger underlying material who took six minutes to ramble through it.
Two, the people who improved the most weren't the people I gave the most feedback to. They were the people who did the reps. Five mock interviews in a week beat one mock interview a month with detailed notes. Volume of pressure-tested practice mattered more than any one piece of advice.
Three, I couldn't scale this. Forty-five minutes per session, write-up takes another thirty. I could help maybe ten people a month at most, and they each paid nothing because asking friends-of-friends for $200 felt strange. There was no sustainable way to give every candidate who needed it the same thing.
Why voice AI was the answer
The version of Talentee that exists today started with a question: what if the AI could run the interview? Not generate questions for the candidate to read silently — actually conduct the interview, voice-first, with realistic timing and follow-ups based on what the candidate just said.
Voice matters because the gap I described above is mostly a voice-delivery gap. Reading practice questions in your head doesn't simulate the cognitive load of formulating a coherent answer out loud while a stranger evaluates you. The brain processes spoken composition differently from written composition. You have to practice the thing you're going to do.
So Talentee runs as a voice conversation. You talk, the AI interviewer listens, asks follow-ups that respond to what you actually said, and the session ends roughly when a real interview would. No reading scripts. No typing. The same cognitive pressure you'll face in a real screening.
Then it scores you on the dimensions that actually predict offer outcomes. Structure (did you answer the question that was asked, in a recognizable shape?). Specificity (did you ground claims in details only you would know?). Pacing (did you spend the right amount of time on the right parts?). Confidence (did you commit to your answer or hedge it into nothing?).
The scoring is the part I'm most proud of, because it's the part that closes the feedback loop that was missing from real interviews. After a real screening, you don't know why you got rejected. After a Talentee session, you know exactly which dimensions you're weakest on, and you can run the next session targeting that specific weakness.
Who we built this for
The first beta users were law students preparing for BigLaw OCI — a process where 1,200+ candidates compete for ~600 summer associate offers across the top firms, and almost everyone has the same paper credentials. The interview is what separates offers from no offers. We've since expanded to candidates preparing for consulting case interviews, banking screens, MBA admissions, tech screens, and the standard behavioral interviews that show up in almost every hiring process.
For schools, we offer a B2B version: career centers can give every student on campus unlimited practice for the academic year. That model exists because the gap I described isn't equally distributed. Students at schools with strong, well-resourced career centers get hours of one-on-one mock interview practice. Students at schools without that resource get a packet of tip-sheets. We close that gap by making the practice itself free at the point of use.
What I'm not trying to do
I want to say this part directly because there's a category of "AI interview tool" I won't build and don't compete with.
Talentee does not coach candidates to cheat. We don't ship a real-time hidden assistant that whispers answers in your ear during a real interview. We don't suggest that you should hide your use of AI from interviewers. The category of "AI cheating tools" exists, it's growing, and it's a fundamentally different product — one that erodes trust in the hiring process and harms candidates over the long run by getting them into jobs they can't actually do.
Talentee is the opposite. It's preparation. The work happens before the interview, not during it. You walk in to a real interview without anything in your ear, having done forty practice sessions on the exact question types you're about to face. That's the kind of preparation that's always existed for candidates whose families could afford a $400/hour coach. We're trying to make it available to everyone.
The honest tradeoff
A real human coach is better than Talentee. I want to say that plainly. A human coach who's interviewed thousands of candidates in your specific industry will give you nuanced feedback that the AI can't yet match. The ceiling is higher with a human.
The reason Talentee still wins for most candidates: the human coach costs $300-500 per hour-long session, requires booking, and gives you maybe four sessions before your interview. Talentee gives you fifty sessions for a fraction of one human session's cost, available at 2am the night before your callback. For most candidates preparing for most interviews, fifty repetitions of focused practice beats four hours of premium coaching.
The reps win.
What we're building next
The roadmap that matters most to me right now: post-interview reflection. Right now, Talentee gives you a score and a breakdown of which dimensions to work on. The next version will let you describe a real interview you just had — the question you got, the answer you gave, what felt off — and get the same dimensional analysis on that real performance. The goal is to close the feedback loop on real interviews, not just practice ones.
Beyond that: more interview formats. Case interview structuring for consulting candidates. Live-coding-conversation rehearsal for engineering screens. Negotiation rehearsal for offer-stage conversations. The same voice-first, dimension-scored architecture, applied to every conversation where stakes are high and feedback is scarce.
How to try Talentee
If you have an interview coming up — any kind, any industry — try a free session. It takes about four minutes, doesn't require a credit card, and you'll get an honest score on the dimensions that predict offer outcomes. If the score is what you want, you're already in good shape. If it isn't, you now know what to work on.
The goal of this company is simple. The candidate who can do the job should get the job. Talentee is what I'm building to close the gap between those two things.
— Reda Fettah, founder of Talentee