About

We rebuilt the screening layer.

Hiring is broken between job posting and offer — the part where 200 applicants become a shortlist of 5. Recruiters drown in CVs. Candidates wait three weeks for a first call. Good hires get lost in the noise. We built Raffi so the screening layer scales without screening fatigue.

TL;DR

Raffi is an agentic AI recruiter built by operators who’d hired hundreds of people and seen the funnel break in the same place every time — the screening layer between job posting and offer. Our thesis: every applicant deserves a real interview, every hire decision should be evidence-based, and no one should pay a placement fee. We’re building the company we wished existed when we were the ones drowning in 200-applicant inboxes.

The screening layer matters because the recruiter-screening phase consumes 25-40% of total time-to-hire (Aptitude Research) and 60% of recruiter time per LinkedIn’s Future of Recruiting report. The cost of a bad hire averages 30% of first-year compensation (SHRM) — almost always traceable to a rushed first-round screen. Fixing screening fixes everything downstream.

What we believe.

Every applicant deserves a real interview.Not a 30-second resume skim. Not a “we’ll be in touch.” A real conversation, in their language, on their schedule. We built Raffi so giving every applicant a 10–15 minute interview is logistically possible — even when you get 500 applicants per role.

Hiring decisions should be evidence-based, not memory-based.Recall bias is real. By the 10th candidate you can’t remember the 3rd. Raffi gives you full transcripts, rubric-anchored scores, and side-by-side comparison — so when you decide, you decide on what was actually said, not what you think you remember.

AI transparency is non-negotiable.Every Raffi call opens with the candidate being told they’re talking to AI. We A/B tested transparency vs. ambiguity. Transparency wins — candidates are more relaxed and more honest when they know what’s on the other end of the line.

You should never pay a placement fee.Raffi is SaaS. You pay your plan, you pay for the actions Raffi takes (calls, reveals, invites). When you hire, you don’t pay an extra dime. The placement-fee model exists because recruiters have leverage at the moment of offer — we removed that leverage by being on the screening side, not the candidate-ownership side.

What we don’t do.

We don’t replace your recruiter. We don’t make the final hire decision. We don’t do live technical interviews — that’s better suited to a take-home or a second-round human. Raffi is the layer between “200 applicants” and “5 great candidates to meet” — not the layer above or below that.

Built by hiring nerds.

Raffi is built by a team that’s hired hundreds of people across sales, engineering, design, and operations — and been on the other side of as many interviews. We know which questions surface signal, which steps slow the funnel, and which excuses kill good hires. The product reflects all of that.

The category we’re building in.

Agentic AI recruiting.Not a chatbot, not a one-way video form, not a résumé parser. An agent that runs the screening loop autonomously — generates the JD, distributes the role, calls every applicant, scores them, ranks them, schedules the finals. The category is <3 years old; the first cohort of buyers (mostly RPOs and high-growth scale-ups) reduced time-to-hire by 30-50% in the first quarter on the new stack per Aptitude Research and Korn Ferry benchmarks. The economics are getting hard to ignore — but the trust layer (anti-cheat, transparency, audit trails, fairness) is still under-built across the category. That’s where we’re investing the most.

How we think about fairness.

Every Raffi interview opens with the candidate being told they’re talking to AI. Every job uses a structured rubric set before the first interview runs, so every candidate is asked the same questions and scored on the same criteria. Anti-cheat scoring catches AI coaching without disqualifying anyone — it flags risk so the hiring manager can decide. We run quarterly bias audits against the rubric and publish a Responsible AI page describing what we do and don’t use the model for. That’s the baseline; the legal layers (EEOC Title VII guidance, NYC Local Law 144, EU AI Act) require all of this anyway, and we treat the floor as the starting point, not the goal.

Why no hire fees.

Placement-fee recruiting exists because external recruiters control candidate ownership at the moment of offer; that’s where their leverage sits. We deliberately built Raffi one layer earlier — on the screening side, not the candidate-ownership side — so we never need that leverage. The candidate is yours the moment you reveal them. You can hire them, pass on them, or ghost them, and no money flows to us. That’s the SaaS promise, and it’s permanent.

See Raffi work

Try Raffi on a fictional role.

5-minute demo interview. No signup. No card. Experience exactly what your applicants will.

Try Raffi yourself

Frequently asked questions

Who built Raffi?
Soltane Yousfi (founder) and a small team of operators who'd hired hundreds of people across sales, engineering, design, and ops — and been on the other side of as many interviews. We started Raffi after running into the same broken funnel one too many times: 200 applicants, 30 phone screens, a recruiter ghosting the candidate at week three, the role still open at month two. Raffi is what we wished existed.
What is Raffi's mission?
To make every applicant interviewable. Hiring is the most consequential decision a team makes, and it's currently optimized for filtering people out (résumé skim, ATS keyword match) instead of finding signal. Raffi flips that — every applicant gets a real 10-15 minute interview, in their language, on their schedule. The right candidate stops slipping through the noise.
How is Raffi different from HireVue, Paradox, or Metaview?
Three differences. (1) Conversational, not async-form — Raffi runs a real voice interview, candidates complete at 95%+ vs the 60% completion rate Talent Board reports for one-way video. (2) SaaS pricing, not per-seat — you pay per action ($0.45/min call, $0.30 reveal), no annual lock-in. (3) Anti-cheat scoring built in, not as an add-on — every transcript ships with a risk score on AI coaching, off-camera prompts, and answer-latency anomalies.
Where is Raffi based?
Distributed team across EMEA and APAC. The product is run as a US-Eastern + Singapore overlap so we cover most timezones. Customer support runs on a 12-hour shift rotation, with most tickets answered inside the hour during business hours.
Why is Raffi a 'he'?
Raffi is a he. We chose a name and a pronoun early because faceless AI agents are harder to trust, harder to give feedback to, and harder for hiring managers to talk about. Naming him made the product better — candidates relax faster on the call, hiring managers reason about him more like a teammate. The brand voice everywhere on this site reflects that.
What does Raffi NOT do?
We don't replace your recruiter — Raffi is the screening layer, not the strategy layer or the offer-negotiation layer. We don't make the final hire decision. We don't run live technical interviews — those still belong to a human or a take-home. And we don't take a placement fee, ever — the candidate is yours the moment you reveal them.