Boardvision turns your PGNs into an evidence-weighted skill map, then builds sparring, review, and transfer checks around what you actually need—not what is easiest to count.
The first alpha milestone is focused on PGN diagnosis. Adaptive sparring and curriculum follow only after their quality gates pass.
Stockfish-bounded
Prediction before reveal
Skill ≠ style
Transfer over streaks
The method / 01
The adaptive opponent is one part of the system. The real product is a learning loop that returns evidence to your profile and asks whether the lesson survives a new context.
Import your games. Boardvision separates skill, style, and uncertainty instead of collapsing them into one mystery score.
Play serious games whose critical decisions sit near your current boundary, always inside hard objective-quality gates.
Recall candidates, commit to a line, and explain your thinking before any corrective answer is revealed.
Return to the concept on a spaced schedule, in varied positions that do not announce what is being tested.
Use unfamiliar neutral positions to distinguish durable transfer from recognition and short-term repetition.
The opponent is not the product.
Transfer is.What changes / 02
Boardvision is designed to know the difference between what your games show, what they only suggest, and what the system still needs to test.
01 · Diagnosis
Objective engine analysis is combined with position features, recurring error clusters, game phase, and time context. Historical games without enough evidence are treated as uncertain—not as permission to invent a diagnosis.
What the report keeps separate
No single-number personality label.
02 · Sparring
Stockfish remains the objective oracle. Personalization chooses only among moves that pass hard quality gates, shaping the decisions you face without discarding chess truth.
Planned session modes
Quality gate · always onUnbiased measurement; no personalized target data.
Serious play with broad, mild personalization.
Selected weak skills, varied across real contexts.
Practice against styles that expose recurring trouble.
03 · Review
At critical positions, Boardvision asks what you were trying to do, which candidates you considered, what line you expect, and how confident you are. Only then does corrective feedback appear.
Critical-position review
04 · Transfer
Exact recall is not enough. Boardvision is designed to check the same skill in a new move order, pawn structure, phase, and eventually an unprompted serious game.
Evidence ladder
After delay
New position
New context
No prompt
Maintenance begins only after evidence holds beyond the original item.
Designed against shortcuts / 03
More games and more solved positions are not proof of durable improvement.
Preferring complications is a tendency. Mishandling them is a skill signal. They remain separate.
A guessed move is not the same evidence as a move supported by reconstructed reasoning.
A familiar answer does not mark mastery until the idea survives unfamiliar contexts.
Private alpha / Current milestone
Boardvision is being built as a sequence of testable milestones. The first private-alpha slice centers on diagnosis that can show its evidence and its uncertainty.
Opens your email app. There is no automatic waitlist, account creation, or form submission on this site.
First alpha slice
After quality gates