Everything you'd ask before you apply.

Miatz is a selective, founder-led engineering institute that trains T-shaped, AI-native engineers — people who can architect, build, operate and own entire digital systems using AI, and prove it with an auditable record. It's . Read the manifesto →

MEE-ahts. The idea: we demystify software, and you leave with 'Miatz' — hard-won judgment worth paying for.

Aspiring and early-career engineers who want real judgment, not another certificate — and companies who want to hire or train them. For students →

Miatz Chuun (say: choon) — our placement test. You start free; Chuun places you, measuring reasoning, product sense and attention to detail, not syntax recall. Cohort seats stay selective on purpose.

Yes — founding cohorts pay nothing: no fees, no income-share, no card on file. We want proof and community before pricing. Why it's free →

No. The Chuun measures how you think, not your résumé. We've built the curriculum to take a motivated beginner all the way.

A daily loop: spaced-repetition recall, a focused chapter, one coding rep, a journal — plus a Saturday review ritual, mastery gates you must clear, five AI labs, and War-Room incident drills. The full method →

Everything you study, journal and decide is indexed into a personal RAG. The AI tutor answers grounded in your own history — you study on your own memory.

Xook is your AI shadow engineer: it works propose-and-approve on real tasks, explains in context, and logs every decision to your auditable record. Meet Xook →

Incident-response drills built on real, documented industry outages. You run the call — diagnose, mitigate, communicate, write the postmortem — and get graded. See War Rooms →

A cryptographically signed proof-of-judgment credential, backed by an auditable record of your real decisions — verifiable by any employer in one click. About the credential →

Yes — you hire engineers with a demonstrated judgment record, not a certificate. You can also train your own team on our engine. Hire from us →

Yes. The same engine grooms corporate teams onto any new product, migration or topic, with Xook shadowing and readiness telemetry. Miatz for corporates →

A full refund inside the first two sessions or seven days, whichever is longer — no forms beyond a request.

Every track has built-in catch-up mechanics and a streak-repair grace window — falling behind pauses progress, it doesn't erase it, and re-entry into the next cohort is always open.

A published daily time-box, not an open-ended "a few hours a week", tied to short, purposeful touchpoints across the day — designed to fit into a day rather than require one.

No. Every session is recorded for life, the curriculum is self-paced underneath the live layer, and lessons download for offline use.

Lessons, quizzes, and reflections all work offline and sync the moment you're back online — nothing is lost or has to be redone.

The placement test places you — it doesn't gate you. The foundations track assumes no prior AI experience and starts with judgment and fundamentals, not a coding bar.

Each level is gated by passing the level before it, not a vague "intermediate" label, so the prerequisite is always a specific, named credential.

No — tracks exist across roles. The core skill Miatz teaches is judgment about AI systems, which extends well beyond engineering.

Yes — pricing is per-level (pay for depth, not a flat bundle), India-first prices are the default where they apply, and every purchase comes with reimbursement-ready invoicing your finance team can process.

Entry tiers run on sandboxed platform keys with hard budget caps, so nobody pays a model bill to get started. When you're ready, you bring your own key (BYOK) and watch exactly what each run costs, in real time — no markup, no surprise line item.

None. Miatz is deliberately model- and vendor-agnostic — you learn transferable judgment, not one company's interface, and sandboxes come pre-provisioned.

No — hard caps at the sandbox tier, a per-run cost meter once you're on your own key, and a team-level spend view for anyone managing seats.

Tuukul (Yucatec Maya: thought) is the structured-thinking strand of Miatz AI-Fluency. It trains judgment at the three gates of AI work — Delegate, Accept, Deploy — with a three-minute daily check-in, drills on your real backlog, and a personal calibration curve computed from your actual outcomes. One score (the Itz'at Score), one credential.

No — structurally. Raw prompt and response text never leaves your machine: the upload schema has no field for it and rejects raw-text-shaped keys. Only judgment signals sync (counts, flags, your own 240-character predictions), you can pause or erase everything, and organizations only ever see aggregates of five or more people.

Yes — that's the Judgment at Scale program: in-flow coaching for your people, and honest aggregate readiness dashboards for you (never individual surveillance). We're onboarding 6–8 founding design partners; see /business/fluency.

Your content stays yours by contract. Miatz is licensed to ingest and score it to build the academy, never to own or resell it, and regenerated material is assigned back to you.

They keep verifying. The credential is a standards-based, signed artifact your own domain can anchor, independent of an ongoing subscription.

Yes. Itz'at credentials are signed, standards-based artifacts that verify independently, without calling Miatz's servers. If Miatz vanished tomorrow, every credential already issued would still check out.

Yes, on a defined wait-and-retake schedule at reduced cost. Failing an eval-gated exam is expected sometimes — that's what makes passing one mean something.