For firms
The recruiting standard
for the AI age.
Granular helps firms find the best talent in the world for any role. It screens applicants and searches beyond the pool, finding the evidence a CV usually misses: dated Granular profiles, transcripts, projects, talks, clubs, public artifacts, and broader proof of work. Then it ranks the strongest candidates with citations.
What follows is a representative graduate hedge-fund pool, built to show how granular reads candidates end to end. Press play.
The layer that has been missing
A granular profile is a record of moments at the desk. The question she put to Claude one Tuesday in March, working through a derivation that did not behave. A debug trail at two in the morning. A point where she disagreed with the model and turned out to be right. The work was always there, because the AI was in the room. The candidate decides which moments to publish.
A granular profile is dated, not retroactive
Every entry on a granular profile is dated to the AI session it came from. The candidate can decline a moment, or redact one, but cannot move the dating, and cannot fabricate a session that did not occur. The result is a different kind of evidence than a CV: derivations from before the paper was published, questions chased before the press caught up, reasoning composed at 2am on a Saturday, all of which are difficult to manufacture in a way the dating would survive.
An agent that reads a pool the way you would
Drop the candidate folder, name the role. Granular adds the working profile the candidate kept while the work was happening, pulls grades line by line from the actual transcript, looks up every named club against its alumni and competition record, opens every named project and traces the contribution history. Returns a ranked pool with citations. Where a signal is missing, the agent says so.
What the agent does, in detail
- 01
Ingests the pool.
Drop a OneDrive, Google Drive, or Greenhouse export. Granular reads every CV and parses every named course, claimed result, and cited affiliation.
- 02
Pulls the layers the CV doesn't carry.
For each candidate granular fetches the LinkedIn timeline, the university transcript with per-course grades and graded coursework (not the GPA line on the CV), and the granular working profile if the candidate has been keeping one.
- 03
Cross-references the layers.
Every claim on the CV is checked against every other layer. A claimed officership the club's archives don't list. A claimed grade not matched in the transcript. Granular flags it for the human to decide.
- 04
Reads what the candidate names.
Granular looks up every named club: its alumni, archives, competition record. Pulls every named talk and reads the Q&A. Opens every named project, traces the contribution history, judges the technical level on the actual code.
- 05
Names the standouts.
Granular scores each candidate across six traits: reasoning quality, originality, working habits, intellectual honesty, calibration under uncertainty, communication under pressure. The ranking returns with per-trait scores and the evidence each is built from. You read it in under a minute, not a week.
For firms · early access
We are onboarding a small number of firms in 2026.
Granular for firms is free for the first three searches. To run a search on your own pool, apply below. A member of granular's team replies within two business days.