All use cases

Resume Screening Pipeline

Rank a folder of resumes against your job spec.

PipelinesAgent: claude
Integrated Services
OpenAI
Required Configuration

This use case works best with the following environment variables, which are injected into your sandbox at runtime:

OPENAI_API_KEY
Prompt
Build a resume screening pipeline for a hiring manager: 1. A job spec editor: role, must-have criteria, nice-to-haves 2. Bulk upload of resumes (PDF or text; include 6 realistic sample resumes of varying fit) 3. Extract text server-side and score each candidate against the criteria using the OpenAI API — every score must cite the exact resume lines that justify it 4. Ranked review board: score breakdown per criterion, cited evidence, and a red-flag column (gaps, missing must-haves) 5. Shortlist export with reasoning included — the hiring manager owns the decision, the tool only organizes evidence Use structured JSON output from the model. Show cost per screening batch. Frame all AI output as advisory, never auto-reject.
Use this

Why this exists

A hundred resumes for one role is a reading problem, not a judgment problem, the judgment takes minutes once the reading is organized. Most AI screening tools get this backwards: they hide the reading and the judgment behind a score you can't interrogate. This blueprint takes the opposite stance: organize the evidence, cite everything, and leave the decision where it belongs, with you.

What the agent builds

A screening pipeline with a job spec editor (role, must-haves, nice-to-haves), bulk resume upload, and AI scoring where every score must cite the exact resume lines that justify it. The ranked review board shows the breakdown per criterion, the cited evidence, and a red-flag column for gaps and missing must-haves. Shortlist export includes the reasoning. Nothing auto-rejects; all AI output is framed as advisory.

It ships with six realistic sample resumes of varying fit, so you can watch how the scoring behaves, including where it hedges, before uploading real candidates.

Make it yours

  • "Add a comparison view for my top three candidates, criterion by criterion."
  • "Anonymize names and photos during the first screening pass."
  • "Track candidates through stages: screened, phone call, onsite, offer."

FAQ

Is this compliant to use for hiring? The tool is deliberately advisory-only, it organizes evidence and never auto-rejects, which is the posture responsible-AI hiring guidance points toward. Your hiring process and jurisdiction's rules still apply; the audit trail of cited evidence helps you meet them.

What does it cost to run? The dashboard shows cost per screening batch. Screening uses your own OpenAI API key, typically cents per batch, not per-seat SaaS pricing.

Can candidates' data stay private? Resumes are processed in your sandbox with your API key. No third-party recruiting platform sees them.

Resume Screening Pipeline | OutcomeDev Use Cases