Every AI study tool eventually faces the same request: just give me the answer. It is the easiest feature in the world to ship and the most corrosive. We want to be plain about where we stand, because the whole product is built on this line.
The deal on offer from answer-machines looks great and is terrible. You trade the struggle — the only part of an assignment that changes you — for the artifact, which changes nothing. The problem set was never the point. The version of you who could do the problem set was the point. Skip the first and you simply do not get the second, and no transcript line will cover for that in an interview, a lab, or the exam room where the chatbot cannot follow.
Incentives, not lectures
We are not interested in moralizing at students; we are students. The honest framing is incentives. Tools that print answers are optimized for tonight’s deadline. Learning compounds on a different timescale, and a tool that actually serves you has to be loyal to the longer one — even when that is less satisfying at 1 a.m.
What that means in the product
- The tutor guides instead of solving. In Socratic mode it asks the question that gets you to the answer; it will not shortcut the arrival.
- Everything is grounded in your course. Answers cite the slide or page they came from, so you can check the source instead of trusting a vibe.
- TeachBack inverts the desk entirely: you explain, and Sapling’s job is to find the step you skipped.
- Mastery moves on demonstrated understanding — quiz performance, explanations, recall over time. Reading something twice does not make the graph greener.
None of this makes us the fun tool on the night before a deadline. We can live with that. The bet behind Sapling is that understanding compounds, that students actually want it, and that software can be on the side of the longer timescale without pretending the deadline does not exist. Guidance over answers. Process over shortcuts. That is the deal we think is worth taking.