Miatz mentors coach a hand-picked cohort on an auditable record: every answer, prompt, commit and incident call is readable, reviewable and attributable. You bring the standard. The console that carries it — the caseload, the learner page, the review queue — is built and not yet live, and every card below says which is which.
Mentoring usually means repeating yourself to people who did not read the docs. Here the gate filters first, the AI absorbs the repetition, and what reaches you is a decision worth arguing about.
Chuun is the free entrance sitting, so you coach a small group that already did the homework. Depth over headcount — the reason good engineers stop mentoring is the reason we keep cohorts small.
Exam answers with written reasoning, coding submissions with history, prompts, incident-drill transcripts, the agent actions a learner approved — the whole record of how they decided, in one place.
The AI shadow handles the tireless one-on-one explaining and flags exactly where a human should step in, so your hours land on judgment instead of syntax questions.
Mentor sign-off is one of the inputs a learner's Itz'at encodes. Your standard becomes part of a credential an employer can verify — attributed, on the record.
Nobody needs another syntax tutorial. What compounds is how you scope a change, argue a trade-off, and decide under an incident clock. That is the whole curriculum here.
No slide decks to build, no LMS to wrestle. Bring your review instincts; the platform brings the content, the reps and the record.
Mentoring is a compensated role, not a volunteer one. Terms are agreed per engagement, because the commitments differ — a review queue, a live session series and a capstone panel are not the same job. We do not publish a rate card or a revenue share, we do not promise a volume of learners, and nothing here is an earnings projection. You will have the terms in writing before you take a cohort.
A mentor here reviews real work and writes notes. They do not grade an exam and they do not certify anyone — that is a separate module, and saying so is cheaper than letting you find out. Everything below is live as of the 2026-09-19 release. Beta rather than Available because none of it has carried a real caseload yet: nobody has been assigned, so every screen is honest and empty until someone is.
You are assigned to named learners — one at a time, a whole cohort at once, or by a rule — and the assignment is what opens their work to you. End it and the reach ends on the next request, while the row and its dates stay. Your organisation sets a ceiling on how many learners one mentor holds, and an assignment past it is refused where it happens rather than quietly dropped.
Their progress, their recent work, the questions they wrote in their own reflections, and the notes written about them. Open a learner you are not assigned to and there is nothing to open — not a polite refusal over a real page, nothing that resolves at all.
Everything waiting on a reviewer arrives in one place, nearest deadline first. You hold three items at a time, so work gets finished instead of collected. Nobody is handed their own work and nobody can resolve it — both are constraints in the database, not buttons kept out of sight.
Each item records when it should have been picked up. Passing that marks it late and shows it. Nothing is auto-reassigned and nothing is dropped: a queue that quietly reorganises itself is one nobody can reason about, and the point is that a pile-up is visible while it is still small.
Every note is written either for the learner or for staff only, and that tier is applied every time it is read. Notes are append-only — a correction is a new note — so what a mentor understood at the time survives knowing how it turned out.
A question a learner writes in a daily reflection surfaces on their page instead of waiting to be asked again. Whether it has been answered is inferred from a later learner-visible note, not known: an answer you gave on a call leaves the question open, and the page counts the ones an unrelated note may have hidden.
Why not Beta: Beta on a public page means live and early. Every screen above is written, reviewed and waiting on one database change that has not been applied to production, so a mentor opening it today would find nothing at all. Coming soon is the honest word, and this page is the first place it changes.
Miatz Certify runs an integrity ladder, and the top of that ladder needs humans. Both roles below are in development — we are naming them now so mentors know where the bench leads, not to promise anyone a seat.
Miatz Kanan is our proctoring layer (kanan — Yucatec Maya, to guard). Its live tier needs human proctors on a console: real-time session review, incident flagging, session attestations. It is intended as a paid role, and when it opens we will publish the bar and take applications from everyone.
Miatz T'aan is the AI oral examination (t'aan — speech). Borderline results go to human moderation: read the transcript, weigh the rubric, make the call. Also intended as a paid role for experienced mentors.
Straight about status: Miatz Kanan and Miatz T'aan are not live yet, so neither role is open and no timeline is promised. When they ship we will publish the bar and the application route publicly. Mentoring here is not a queue ticket for a future job — it is worth doing on its own terms.
We read every note. If your standard fits, we aim to have you reviewing within a couple of weeks.
One note: your domain, the kind of engineer you'd sharpen, and roughly how many hours a week you can hold.
We talk through how you review work and where your standard is uncompromising. Thirty minutes, usually one call.
You read a live cohort's record and review alongside an existing mentor, so you see the work before you commit.
You get a cohort, a direct line to us, and the console the moment it is live. Scope grows only as fast as you want it to.
Include your domain, a link to something you have built or reviewed, and the hours you can hold in a week. That is enough for us to reply properly.
Most mentors hold a few hours a week: a review pass over the cohort's submissions plus a live session. Xook is there specifically so those hours go to judgment calls rather than repetitive explaining.
Not yet, and that is why every card in it is chipped Coming soon rather than Beta. The console is built and reviewed — the caseload, the learner page, the review queue, the notes — and the database change it depends on has not been applied to production, so none of it is running for anyone. Beta would tell you that you could use it today. When it is applied, the chips on this page are the first thing that changes.
Their own work: progress, submissions, reflections, check-ins and plan. The database decides this rather than the screen, so there is no view of a learner you are not assigned to anywhere, including through the API. Two things assignment does not grant: you cannot change a learner's progress, and you cannot mark their work complete.
No. A mentor reviews real work and writes notes. Resolving a review records a judgment about one piece of work and leaves a trace of it; an Itz'at is earned when mastery gates clear on the record, in a separate module, and no single person can hand one out.
No. The curriculum, daily reps, exams and drills already exist on the platform. You bring review instincts and standards; you can propose content changes, but nothing about mentoring requires authoring a course.
Yes — mentoring is a compensated role. Terms are agreed per engagement rather than published as a rate card, because the scope varies: a review queue, a live session series and a capstone panel are different commitments. We do not publish a revenue share, we do not promise a volume of learners, and nothing we discuss is an earnings projection — you will have the terms in writing before you take a cohort. Miatz Kanan proctoring and Miatz T'aan moderation are also intended as paid roles, and neither is open yet.
Miatz Kanan is our proctoring layer. Its live tier is staffed by human proctors working a session console. It is in development, not open yet — when it ships we intend to open a certified-proctor path as a paid role, and we will publish the bar and open applications to everyone at that point.
Not yet. Miatz T'aan is the AI oral viva; human moderation of borderline results is planned as a paid role once the surface is live. We will publish the bar and open applications then rather than pre-committing anyone.
Yes — that is the common case. Reviews are asynchronous; only the live session needs a fixed slot, and we schedule it around you.
Not alone. An Itz'at is earned when mastery gates clear on the record — exams, verified skills, solved challenges, drill scores — with mentor sign-off as one input. No single person can hand one out, and no one can talk their way past a gate.
From the guides: how cohorts are run, what the learner 360 shows, how review queues work, and the mastery gates behind every advancement.
Create cohorts, review applicants and enroll learners — with seat quotas enforced automatically.
Read the guideOne page per learner: vitals, integrity, AI usage, every artifact they produced — and the levers to coach them.
Read the guideWhat lands in each review queue, and what happens when you action it.
Read the guideHow exams are graded and how you advance.
Read the guideBring your review instincts to a cohort that already cleared the gate — and to a credential your sign-off helps make real.