Hiring is the only thing a company does that starts over every time.
Every other function compounds. Code gets refactored, customers get understood, processes get better. Hiring doesn’t. The last search taught you something and it left with the person who ran it. The bar lives in someone’s head. The candidate you passed on in March is gone by June, and nobody knows why you passed.
Most hiring tools don’t fix this. They’re filing cabinets — they store what someone typed in and hand it back. The knowledge still leaks out the sides.
So we built our own, around the four things that have to be true for hiring to get better instead of resetting.
Everyone means something different by “strong.”
Most hiring goes wrong here, quietly, before a single résumé is read. Nobody writes the bar down, so everyone interviews against their own — and six weeks later a good candidate is stuck between two people who never agreed what they were looking for.
- The bar, written down
- What this person owns, what great looks like, and what’s failed here before — written down as a scorecard, not left as a vibe in someone’s head.
- The role, defined properly
- Title, description, and screening questions built from the real requirement — so the posting reflects this job, not the last one that looked like it.
- Scored against this role, not roles in general
- Ten dimensions, each one specific to what you’re hiring for — never a generic template stretched to cover every role.
- Nothing skipped
- The process is enforced by the system, not by discipline — stages can’t be jumped or quietly dropped when someone’s in a hurry.
The field, not the funnel.
Applications tell you who happened to be looking, in the fortnight your post was up. The person you actually want is busy, employed, and not reading job boards. We’ve spent years building a pool of people worth knowing, and we add to it every day — so a search doesn’t begin with a search.
- Ranked before you ask
- Open a role and the best matches from the pool are already ranked and waiting. The search runs before anyone types anything.
- Asked for the way you’d say it
- “Java developers with fintech experience, five to eight years, in Austin.” It reads for meaning, so someone whose résumé says “payments infrastructure” surfaces without the word ever appearing.
- Every referral tracked and credited
- Partners submit people, follow them through, and always get the credit. The best people still come from someone who knows them.
- Nothing sits in an inbox
- Applications are pulled in, read, and scored before a recruiter has opened their mail.
What we learn about someone shouldn’t leave with whoever learned it.
In most processes, everything real about a candidate lives in one recruiter’s head and one interviewer’s memory. So the fourth interviewer asks what the first already answered, the good reason someone was passed over is lost by the next quarter, and the whole thing has to be reconstructed from scratch.
- Every résumé read properly, and read sceptically
- Not transcribed. Read — including the parts that don’t add up. Titles inflated a level, timelines that don’t reconcile, skills claimed with nothing behind them.
- Judgment with its workings shown
- Every score says where the evidence came from and how sure it is. There’s a difference between “this is a weak candidate” and “there isn’t enough here to judge.”
- Weak spots don’t get averaged away
- One serious gap can’t hide behind nine strong areas.
- Nobody walks into an interview cold
- Every interviewer gets the candidate, what earlier rounds flagged, and what to probe — before they sit down.
The fifth hire is better than the first.
This is the part that doesn’t exist anywhere else. Every hire teaches the system something about how you hire — what you say yes to, what you pass on, and why — and none of it leaves when someone changes job.
- It changes its mind
- When interview evidence lands, earlier assessments get revised against what people actually observed — and it records that it changed its mind, so you can see where the first read was wrong.
- It learns which sources actually work
- Not which produced the most candidates. Which produced the people you hired and kept.
- It grades its own homework
- The system measures whether its own scoring predicted who got hired. If its judgment isn’t any good, that shows up as a number rather than an opinion.
- Your context stays yours
- Every hire deepens what we know about your bar, your team, and what works here. By the fifth role we’re not guessing.
It has an opinion. It never gets a vote.
Everything on this page is a machine reading, scoring, drafting, and flagging — and then stopping. It never moves a candidate, sends a message, or makes a call. That’s a line we drew on purpose, not a feature we haven’t built yet.
It proposes.
You decide.
When you know better than the filter, the filter loses. Nothing is written to the record until a person accepts it.
Your pipeline can’t touch another client’s. Every change is logged — who, what, before, after — and audit records are never deleted.
Rejections come from a person, by email — never a messaging app. Everyone you don’t hire still speaks well of your company.
See it on a real role.
The fastest way to understand any of this is to watch it work on a role you’re actually trying to fill. Tell us what you’re hiring for, and we’ll show you.
Our team runs the platform. You’ll see everything it sees on your roles.