The Audit Log Is the New Resume

A resume is a set of claims. A certificate is a claim about claims — an institution asserting that, at some point, you passed something. In a world where AI can generate a polished resume in seconds and course certificates are earned by letting videos play in a background tab, both signals are approaching zero.

Employers know this. It's why hiring processes keep adding rounds: nobody trusts the paper, so everyone re-tests from scratch. The signal collapsed; the interview gauntlet is the workaround.

We think the fix isn't a better certificate. It's a different kind of artifact entirely.

Don't certify that judgment happened somewhere out of sight — record the judgment itself, and make the record verifiable.

What a decision record looks like

Every miatz learner works with [Xook](/xook), our AI shadow engineer, in a propose-and-approve loop. Xook proposes a change, a fix, a design; the learner approves, rejects, or amends it — with reasoning. Every one of those decisions is captured: what was proposed, what the learner decided, why, and what happened next. War Room incident runs, code reps, and review decisions land in the same log.

Do this daily for months and you accumulate something no resume can fake: a longitudinal record of engineering judgment. Not 'proficient in databases' — an actual entry where you rejected an AI-proposed migration because it locked a hot table, with your reasoning attached and the outcome recorded.

The Itz'at is the credential built on top of that log: a cryptographically signed, employer-verifiable proof that the record is authentic, unedited, and yours. The name is deliberate — Itz'at, from the Maya word for sage. Not 'attended'. Not 'completed'. Judged, on the record.

How an employer reads one

A certificate is read in one second: present or absent. An Itz'at export is read like a case file, and that's the point. In a few minutes, a hiring manager can see things interviews struggle to surface:

• Judgment under ambiguity. Which AI proposals did the candidate wave through, and which made them stop? The rejections are the interesting part — rejecting plausible-but-wrong output is precisely the skill the AI era pays for.
• Reasoning quality. Every decision carries a why. You can watch reasoning mature over months — early entries citing rules, later ones citing consequences and trade-offs.
• Behavior under pressure. War Room entries show triage order, hypothesis discipline, and what the candidate wrote to stakeholders mid-incident, graded like a staff SRE would grade it.
• Honesty. The log contains mistakes — approved proposals that later broke, hypotheses that were wrong. A record with no errors is a record of no risk. Employers tell us the recovered mistakes are the most convincing pages.

Verification is one click: the signature proves the log came from miatz's systems and hasn't been edited. No transcript office, no 'trust me'.

The interview conversation changes

The best consequence isn't skipping interviews — it's upgrading them. Instead of reversing a linked list on a whiteboard, the conversation becomes: 'Walk me through this decision from March. What would have happened if you'd approved it?' That's an interview about the actual job — and the candidate has literal receipts.

For candidates, this rebalances power. Career-switchers and people without brand-name employers are the biggest winners: the log doesn't care where you worked before; it only shows how you decide. For [companies](/for-companies), it collapses risk on exactly the axis certificates never covered: not 'did they learn the material' but 'can they be trusted with the merge button'.

The obvious objections

Can't you game it? Volume, no — the log is generated by doing supervised, graded work over months inside the program, not by self-report. Quality, also no — decisions are graded against outcomes and mentor review, and the grade travels with the entry.

Isn't it surveillance? It's your record. Learners own their logs, control exports, and choose what to share with whom — the same architecture that powers the [personal memory engine](/glossary/retrieval-augmented-generation) powers the credential. The difference between surveillance and a portfolio is who holds the keys.

Where this goes

Software already trusts logs over claims — that's what an audit trail is for. Hiring is simply the last place still running on unverifiable assertions. We think that ends this decade, and we'd rather build the standard than wait for it.

Your resume says what you are. Your audit log shows what you did. If you'd rather be holding the second kind of document, [start building one](/program).