Nobody gets skimmed.

how the system behaves, not a benchmark

Screen Smarter, Not Harder.

Skip the endless resume pile. Every application is parsed and scored against the criteria you set, and every applicant answers the same short set of questions, so the pile arrives already read.

What happens to an application.

• 01
  The resume is read, not scanned

  A PDF goes through text recognition and is then turned into structured fields: name, contact, summary, skills, one entry per role with its own bullet points, education, and a catch-all for the certifications and projects that never fit a box. Nothing is matched on keywords.

• 02
  Contact details are stripped before scoring

  Email addresses and phone numbers are removed from the copy that reaches the scoring model. What ranks a candidate is what they did, not who they are or how to reach them.

• 03
  Hidden instructions are removed, and kept

  Resumes now arrive with instructions buried in them, aimed at whatever model reads them next. Anything of that kind is excised before scoring, and the excised lines are stored, so you can see exactly what somebody tried.

• 04
  Your questions get asked, word for word

  The questions you set are shown to every applicant exactly as written. There is no rephrasing step and no generated question. Two ship by default, notice period and current versus expected compensation, and you replace them with your own.

• 05
  Answers are scored with the resume, not beside it

  The written answers join the resume as one body of evidence, so a criterion can be satisfied by something a candidate said this week rather than something they wrote last year.

How a criterion gets judged.

Each of your criteria comes back with a verdict, a verbatim quote of no more than thirty words, where that quote was found, a one-line explanation and a confidence figure. The verdict rests on how well the claim is evidenced.

• proven
  A result, a number, a named project, a scope

  “Migrated forty services to Kubernetes and cut deploy time by sixty percent.” Counts in full.

• described
  The work is narrated in context, but nothing came of it on paper

  “Responsible for managing Kubernetes clusters.” Counts, at roughly seventy percent.

• mentioned
  The word appears and nothing else does

  “Skills: Kubernetes.” On its own, counts for nothing. This is the line that separates a real career from a well-prompted one.

• not found
  No evidence, or evidence pointing the other way

  Recorded with high confidence, because knowing something is missing is worth as much as knowing it is there.

What it will not do.

• It will not reject anybody. There is no auto-reject anywhere in the pipeline, at any score.
• It will not read a Word file, a scan or a link. PDF only, for now.
• It will not ask a question you did not write.
• It will not let a resume tell it what to think.