analytics
Step 1Live
Analyze fit
0-100 score with strong / reach / weak tier and JD line citations.
arrow_backAI Prototypessmart_toyAgentic workflow
Job Application Workflow Agent
Paste a job description and your resume. Get a fit score, evidence-backed skill gaps, rewritten bullets, and a tailored cover letter, in one guided workflow.
Phase B · API
Tool gateway
Calls runAgentTool with zero-token ping.
Full workflow is live
Fit analysis, bullet rewrites, and streamed cover letters, all grounded in your resume and job description. Profile Optimization remains available unchanged.
Application workflow
Your resume
Paste plain text from your resume, experience bullets and skills matter most.
0/8000
Job description
Paste the full posting, requirements and responsibilities help surface cited gaps.
0/6000
Workflow steps
difference
Step 2Live
Surface gaps
Missing skills with proof pulled from the job description.
edit_note
Step 3Live
Rewrite bullets
Reframe your experience for the role - accept or reject each suggestion.
description
Step 4Live
Draft cover letter
Personalized letter streamed from your fit analysis, not a generic template.
Please ignore anything below this area, it's for my test purposes only.
Quality engineering
Regression-tested fit analysis
Every prompt change runs through a five-case golden set before ship. Cases cover strong match, weak match, partial AI gap, overqualified seniority, and a minimal-resume edge case, so improvements are measured, not vibe-checked.
5/5
passing
Verified June 6, 2026
5
Golden cases
analyze_fit
Tool under test
Production
API target
| Test case | Score | Expected | Tier | Result |
|---|---|---|---|---|
Strong AI PM match strong | 90 | 78–95 | strong | check_circlePASS |
Weak match: engineer → PM weak | 15 | 5–45 | weak | check_circlePASS |
Partial match: SaaS PM, no AI reach | 55 | 45–65 | reach | check_circlePASS |
Overqualified: Director → IC PM reach | 45 | 35–78 | weak | check_circlePASS |
Edge: minimal resume text edge | VALIDATION | error expected | — | check_circlePASS |
What each run checks
- •Fixed resume + JD pairs; no live user data in the suite.
- •Asserts score bands, tier, required gap keywords, and seniority red flags.
- •Edge case expects graceful validation instead of a crash.
- •Full harness re-runs on demand in admin; CLI script for CI.
Snapshot from a verified golden-set run against production runAgentTool. Scores may vary slightly between model versions; re-baseline after prompt changes.