Practice is live as of this morning. The engine, on demand, in sessions of ten, twenty, or forty-five minutes, against the candidate's current state. No scored band at the end. No test-day frame. Just adaptive sessions a candidate runs whenever the surface fits their day.
The candidates who have been with us since the alpha have been asking for this since spring. We held it until the mastery-vector v2 launch in January because the candidate-routing inside Practice depends on the vector being a real selection input. Without v2, Practice would have been a uniform-difficulty drill set wearing a timer. With v2, it is the engine, focused on the gap the engine has been tracking against the candidate.
What it is
Two choices, in this order. The candidate picks a length — ten, twenty, or forty-five minutes — and a focus. Length is plain wall-clock minutes the candidate can afford. Focus is one of the three sections, a specific sub-skill cluster surfaced from the candidate's gap list, or "Engine's pick," which lets the mastery vector route the session.
Inside the session, the engine runs the same v1.1-with- mastery-routing loop the Mock surface runs in adaptive mode and the diagnostic runs in its banded form. Fisher information against the candidate's current θ, mastery- lift weighted into the selection step, exposure caps, topic-balance constraints. The candidate sees the same item runner — the bookmark mechanic, the timer, the flag-and-review affordance — but the section never commits a score because the session is not a section.
What the session produces
Two things, neither of them a number on a scoring band.
An updated mastery vector. Every response shifts the candidate's vector exactly as it would inside the diagnostic or the mock. The gap list the diagnostic surfaces, the Section Analytics panel currently in alpha, and any future surface that reads the vector all reflect the practice work without the candidate having to log it manually.
A session summary, in plain language. At the end of every session the candidate sees a short readout: which sub-skill the session targeted, which items moved the vector materially, the time-on-task median, and the one item the engine would route to next if the session continued. No scaled-score claim, no predicted-band shift, no "you improved by N points" line. Practice is not a measurement surface.
What we cut
Two features that came up in candidate conversations through the autumn and did not ship.
A predicted-band shift after each session. A bar that says "you went from 615 to 622." We will not ship that. A single twenty-minute Practice session is too small a sample to move a band estimate honestly, and a bar that fakes movement at that granularity is the kind of motivational artefact that erodes trust in the rest of the surface. Section Analytics, when it ships, will show band changes against a fitted window — a week or a fortnight — where the change is statistically defensible.
A streak counter. "You have practiced nine days in a row." We will not ship that either. The engine routes against response density, not against consecutive-day calendar pressure. A candidate who practices three times in a week is in better shape than a candidate who practices once a day every day at five minutes each. We do not want a counter that nudges candidates into the second shape.
How it differs from the diagnostic and the mock
Three surfaces, three jobs.
The diagnostic is a 25-minute banded measurement. It outputs a scaled-score estimate, a sub-skill gap list, and a starting band. A candidate runs it on day one and every thirty days thereafter.
The Mock Exam is a test-day rehearsal. It outputs a per-section scaled score, a percentile readout, and a pacing trace. A candidate runs three across a Pro 4-month window, two across a Pro 1-month.
Practice is the rest of the time. It outputs an updated mastery vector and a plain-language summary of where the session targeted. A candidate runs it whenever they have a quiet twenty minutes and a seat.
Who gets it
All Pro tiers — Monthly, Quarterly, Six-Month — receive Practice immediately. Unlimited sessions, no per-day cap. The free diagnostic now includes three Practice sessions per week at no cost; subsequent sessions move the candidate into Pro 1-month. We added the three free sessions because Practice is the surface that makes the engine legible to a candidate who has not yet committed; a one-shot diagnostic is not enough to feel the routing.
Section Analytics, the surface that reads the vector Practice is updating, is still in alpha and ships in the spring. We will write a longer note when it does.
— Brightroom Product