Write one test note
Record what happened in one bounded state and the method that exposed it. The exercises use a supplied fictional journey.
Public preview · technical self-paced course
Turn one fragile web journey into a reviewable release decision.
A critical web journey can look finished while keyboard focus, state changes, or error feedback break the task. In this technical self-paced course, build one Experience Release Record: a compact record of a named state, the evidence collected for it, its limit, the reviewer boundary, and a bounded PROMOTE or HOLD decision.
Public preview: the practice and XQ.1 are open now. XQ.2–XQ.13 are included in the complete course; enrollment is not available yet.
You leave able to explain and challenge a release decision: which evidence supports it, what it cannot establish, and what must happen next. The record makes uncertainty visible without pretending to settle it.
Build the release record
Place the finding beside its evidence, limit, and owner instead of treating one result as a release verdict.
Human review or HOLD
Carry the record into an existing review, issue tracker, or test notes. A reviewer can challenge it or keep the journey on HOLD; no integration is claimed or required.
Who it is for
Product, frontend, and quality practitioners who own a web journey that must remain usable as it changes. You should be able to inspect basic HTML or browser test output and use a terminal. You do not need accessibility certification expertise, and this course does not provide legal, design, or screen-reader certification.
Before you start
You need a browser, a terminal, and enough HTML vocabulary to read a name, role, state, and selector. No accessibility credential, cloud account, real product, or prior AI-judge experience is a prerequisite. The fictional Library journey and the required no-code practice are the common starting point.
WHAT YOU BUILD
One release record with a traceable shape
Start with a short description of the user goal, critical states, evidence, and owner—your journey contract. Then produce a five-field practice note—the practice packet—twelve predecessor lab envelopes, and an assembled thirteenth record: state, evidence, limit, review boundary, and decision.
HOW THE COURSE WORKS
One fictional journey, 13 ordered modules
The Library reservation journey recurs through eight core modules and five advanced modules. Each new artifact must connect to earlier evidence.
WHAT TRAVELS WITH YOU
A handoff you can adapt
Take the record shape, questions, and evidence vocabulary into your existing release review, issue tracker, or test notes. The course teaches the handoff; you adapt it to your workflow.
Choose your practice route
Everyone starts with the supplied fictional page and the required no-code conceptual practice: inspect it, record one bounded observation, and complete the five-field practice packet. The offline Python labs are the technical extension required only for the builder outcome; they do not replace that shared starting point. The course remains technical: you need the HTML vocabulary and terminal described above. By the end, you have a concrete record to adapt—not a new platform to install.
Make the first five-field observation
Use the supplied fictional page to predict focus, inspect a narrow state, describe what happened, and complete the practice packet: state, method, observation, decision, and limit.
Open the English practice →Assemble the evidence trail
One deterministic Python lab accompanies each module. The capstone assembles the final fictional release record from fresh lab envelopes, then shows the record shape you can carry into an existing workflow.
Start with XQ.1 →Core path · first release record
This is the first eight modules of a 13-module course. They move from the task promise to the first reconciled decision; each includes a worked case, counterexample, no-code attempt, and one builder lab. The remaining five modules are required continuation, not an optional route: complete XQ.9–XQ.13 in order, with XQ.13 serving as the capstone handoff.
PUBLIC PREVIEW BOUNDARY
XQ.1 is open. XQ.2–XQ.13 are shown so you can inspect the complete path, but they open only with the complete course after enrollment is available.
Progress is stored only in this browser.
XQ.1 · CONTRACT
The task that disappears
Turn one user goal, its critical states, evidence, and owner into a journey contract.
XQ.2 · SEMANTICS
A page can look fine and say nothing
Distinguish visible pixels from names, roles, states, landmarks, and messages.
Included in the complete courseXQ.3 · FOCUS
Focus is a path
Record event-to-target transitions through open, error, close, and return.
Included in the complete courseXQ.4 · REFLOW
The small screen is the constraint
Separate acceptable wrapping from overflow, clipping, and task obstruction.
Included in the complete courseXQ.5 · VISUAL CHANGE
A screenshot catches a different defect
Bind a visual difference to baseline, environment, intent, and review.
Included in the complete courseXQ.6 · RUNTIME
A journey needs runtime evidence
Join completed path, browser errors, and terminal capture in one run.
Included in the complete courseXQ.7 · REVIEW LEDGER
The report is not the user
Keep automated coverage separate from human and assistive-technology review.
Included in the complete courseXQ.8 · RECONCILIATION
Release evidence can be challenged
Reconcile seven predecessor records without averaging away a blocker.
Included in the complete courseRequired continuation · richer evidence, same boundary
Modules 9–13 are the required final five modules, not an optional advanced track. Complete them in order: state matrices; a focused contract for the browser accessibility tree—the roles, names, states, and order exposed to assistive technology; risk-to-evidence mapping; and a bounded multimodal evaluator (judge) that compares a screenshot with a visible rubric and only routes cases to human review. XQ.13 is the capstone handoff. The output remains a portable decision record, not a claim that the fictional exercises observe a real journey.
XQ.9
Baseline visual and state matrix
Version visual evidence by state, environment, risk, and review reason.
Included in the complete courseXQ.10
Browser accessibility tree
Make focused role, name, state, and order expectations inspectable.
Included in the complete courseXQ.11
Risk evidence matrix
Match each declared task risk to evidence, owner, and remaining limit.
Included in the complete courseXQ.12
Multimodal evaluator (judge): rubric and calibration
Compare one captured state with a visible rubric, then keep the signal advisory and challengeable.
Included in the complete courseXQ.13
Multimodal capstone: decision and drift
Defend the complete synthetic record through rebaseline and recalibration decisions.
Included in the complete coursedelivery and boundary
Plan for a 13-module self-paced course: local no-code practice, thirteen offline Python labs, and a capstone that assembles a fictional release record. Workload is pending an observed pilot, so no duration is advertised. Delivery is download-based self-study with worked rationales, deterministic output, and first-run email support only. The exercises are synthetic and offline; adapting the record to an existing workflow remains the learner's work.
COURSE ACCESS
A technical course with a visible delivery boundary.
One downloadable, self-paced course: the no-code starting practice, thirteen offline Python labs, worked rationales, and the synthetic capstone record. The price covers the course materials; it does not add a platform, certification, live review, or production integration.
pre-launch priceUS$240
Regular price: US$300
Enrollment is unavailable at this time. Checkout and delivery open after operational validation.