This article is currently available in English.
Most language products are very good at giving you material. A deck can give you another card. A course can give you another unit. A chatbot can give you another conversation.
The harder question is what should happen after you try to use the language.
That attempt contains useful evidence. Maybe the meaning was clear but the form was fragile. Maybe you knew the rule and could not retrieve it quickly. Maybe the exercise was simply too hard for the support you had. A score alone flattens those situations into the same result.
The messy middle
Shaded is for the stretch between recognizing something and using it naturally.
That is why the product starts with real attempts and keeps a learner model underneath the different ways you practice. An answer can update the concepts and skills that need attention. A later activity can bring the same idea back with different support. Progress can remain connected when you move between Flow, Passage, review, goals, and Voyage.
The aim is not to build an endless generator of exercises. It is to make each next exercise more relevant because of what happened in the last one.
Flexible practice needs visible ground
AI makes it possible to shape practice around a learner instead of forcing every learner through the same fixed sequence. That flexibility is valuable, but it should not make the learning material impossible to inspect.
When reliable source material is available, Shaded keeps it attached as an Anchor. A learner can see where an example, dictionary entry, or reviewed item came from instead of being asked to trust an unexplained output.
Anchors do not turn a probabilistic system into a perfect one. They make the supported parts clearer and give the product a better foundation for improvement.
What we’ll share here
Updates will be a working journal, not a publishing quota. We’ll use it for:
- meaningful product releases;
- decisions behind the learning system and interface;
- lessons from building grounded, adaptive practice; and
- company milestones worth recording.
We will favor concrete examples over announcement language. If a change is easier to understand with a screenshot, a tradeoff, or a lesson learned, that context belongs in the update.
Shaded is already available on the web, iOS, and Android. There is much more to make, but the direction is steady: learn from the attempt, keep the evidence connected, and find the next useful step.