About & methodology

How we read your career.

Hedj is a Personal Career OS for knowledge workers in the AI era — a way to hedge the career you have against the model curve ahead of you.

What a reading is

Five axes — Automation, Skill Half-life, Wage Pressure, AI Investment, and Role Demand — each scored 0–100, blended into a single number of months. The number isn't a deadline; it's a market reading. Treat it the way a fund manager treats a valuation: directional, not absolute. Every reading also carries a risk classification (Critical / High / Watch / Stable) and a sensitivity range, because a point estimate without uncertainty is theatre.

Where the numbers come from

Your LinkedIn profile is fetched through CoreSignal (Tier-1 data partner; we don't scrape). The 5 axes are anchored to public reference data:

  • Anthropic Economic Index — automation curves per role family, refreshed quarterly
  • BLS + Indeed — 12-month hiring trend per role
  • Layoffs.fyi — AI-attributed layoff events, rolling 90 days
  • O*NET — task composition for your role family
  • Anthropic Claude— generates the 3–5 questions we ask you, and Sunny's analyst voice
Deterministic, not vibes

The months number is a pure function of your 5-axis scores plus structural modifiers (career level + tenure × level interaction + AI fluency). Same answers, same reading — every time. The LLM picks the question set, grades the axes, and writes the narrative; it doesn't pick your number. The dashboard's "Show the math" card walks every contribution; the full formula is in our research notes.

Structural modifiers, not just axes

On top of the 5 axes we apply three empirically-grounded modifiers. Career level shifts the risk score directly (junior +8, executive −8). Tenure follows an inverted-U for non-executive roles — 0–2 years and 10+ years are both risky in 2026 — and a monotone protective curve only for executives. AI fluency builds the defense bonus that pulls months back up. Why we changed the tenure curve walks the empirical case in detail.

Your data

Anonymous readings work without sign-in. Once you sign in, you own a long-term profile (LinkedIn URL, resume, notes to Sunny) and a history of every scan you've run. Privacy has the full breakdown. You can delete everything from /account/security with one click.

Limits of any reading

The model curve moves faster than the labor market does. Re-read monthly — your axes shift as the AEI refreshes, your skills stack, and your context (the Notes you give Sunny) evolves. Treat a single reading as a snapshot; treat the trend as the actual signal.