Telling Technology · Learning in public
All episodes
EP.27 — GAP MATRIX
Know exactly where your skills stand

My AI read 5 live role postings and mapped my own skills against every one

I pointed it at five real postings from today's market. Before I'd read a line, it had built a Have / Partial / Missing matrix against my own profile and ranked them by fit. A skills radar, not a scroll.

Split scene: five live role postings on the left feeding a glowing Have/Partial/Missing competency matrix on the right.
⌁ 5 postings in · one competency map out
It started with a folder of five postings and one question: where do I actually stand?
01 The raw signal

Five live postings, pulled and stored

Scrolling role boards means guessing — no ranked signal on where your skills actually land. So I gave the market a front door: ingest_jobs.py pulls live postings (reusing my existing scan query and filters) and stores the full text of each, stamped by scan date. Five real postings from today, captured in one run — before I'd read a single word of them.

claude code — desktop
> Pull today's live role postings with the existing scan filters and store the full text of each by date.
Ran ingest_jobs.py — pulled 5 live postings (2026-07-02), stored full text per posting under today's scan date. Deduped against yesterday. Ready to diff against your profile.
Terminal running ingest_jobs.py, pulling and storing five live role postings dated 2026-07-02, company names shown as role types.
ingest_jobs.py · 5 stored
Five live postings captured in one run — company names kept private.
02 Gap-to-me

A matrix that grades every demand against my own profile

Then the real move: cross-reference every competency the postings demand against my own about-me.md profile. Each row lands as Have, Partial, or Missing — and every Partial or Missing row carries a named, specific action to close it. Not a fit score pulled from thin air: a line-by-line map of what the market asks for versus what I can already prove.

claude code — desktop
> Cross-reference every competency these postings demand against my about-me.md profile. Mark each Have / Partial / Missing, with a close-the-gap action per row.
Built the Gap-To-Me matrix. Every demanded competency scored Have / Partial / Missing vs about-me.md, each Partial/Missing row with a named action. Strongest coverage: the Technical PM profile — zero Missing rows.
The Gap-To-Me matrix: demanded competencies marked Have, Partial, or Missing against Jacob's profile, each gap with a close action.
Have · Partial · Missing
Every demand, graded against my own profile — with a fix per gap.
03 The ranked read

Ranked by fit — before I'd read one of them

With every posting graded, the ranking wrote itself. The AI ordered all five by how completely my existing competencies already cover them. #1 came out fully covered — zero Missing rows. The rest sort down from there, each one showing exactly which skills to sharpen to close the distance. A ranked read of where I stand, produced before I'd properly read a single posting.

claude code — desktop
> Rank the five postings by how completely my current competencies cover them. Where's the strongest fit?
Ranked. #1 — a Technical PM profile — zero Missing competencies, strongest fit on the board. #2–#5 sort by rising gap count, each with its top skill-to-sharpen called out. Read the market before reading a word.
Ranked list of five postings by fit, #1 highlighted as zero-gaps, company names redacted to role types.
#1 · zero gaps · redacted
Ranked by fit — company names stripped to role types.
🔒

Names stay private — the method is the story

These are real, live postings — so the companies stay anonymous. Every one shows as a [role type], never a name, and the matrix grades demands against a public profile, not anything private. This isn't about any single employer or going anywhere — it's a compass for where my skills stand against what the market asks for right now. The method is the part worth copying.

04 Four receipts

Ingest → grade → rank → study

Four frames from one run: the postings pulled in, the Have/Partial/Missing grade, the ranked read, and the fifteen study briefs it wrote off the back of it. Tap any image to enlarge it and read the exact prompt that drew it.

BUILT OVERNIGHT · ONE SCAN, ONE MAP

The market, read before I read it

Five live postings went from raw text to a graded, ranked competency map in one run — #1 fully covered, every gap named, fifteen study briefs written before I'd properly opened a single posting. That's not scrolling and guessing. That's knowing exactly where I stand, and exactly where to sharpen next.

Verdict board: #1 role fully covered with zero missing competencies, gaps and study briefs summarized, companies anonymized.
5 postings · 1 competency matrix · 15 study briefs · 0 names published
06 Steal this

Steal the loop: postings → gap matrix → study plan

The pattern is portable — pull the roles you're curious about, grade every demand against your own profile, and let the gaps write your study plan. No private data here; the companies stay anonymous and the method is the part you copy. Everything in this series is free and open.

skill Gap-To-Me matrix script ingest_jobs.py method Have / Partial / Missing + fix output 15 study briefs
try the method Ask your AI: "Grade this role posting against my profile — Have, Partial, Missing, with a fix per gap."

Want the loop pointed at your own profile? Mail jacob@skogstrom.se — one call, no pitch.

Next episode

The map says where I stand. Next: where the market's about to move.

Knowing today's gaps is half of it. The next build points the same engine forward — reading where demand is trending so the study plan gets ahead of the market instead of trailing it.

Darkened teaser of the AIOS engine that ran the competency scan.
drops next · follow so you don't miss it