A new AI tutoring startup, Bloomy, just launched claiming students grew 1.8x faster than expected on a well-known academic growth test. It’s the latest in a string of similar claims from Khanmigo, Synthesis, and LittleLit — and it’s worth asking the same question of all of them.
The claim isn’t the problem. The question it doesn’t answer is.
A rising test score can mean a student built real, transferable understanding. It can also mean a student got faster at the specific pattern of problems that test rewards. Both look identical on a growth chart. Only one of them holds up three weeks later, in a new context, without the app open.
The tell is in the design, not the marketing
Interestingly, Bloomy’s own product requires students to pass an unaided, no-help checkpoint — about 90% correct, alone — before moving to the next skill. That’s the same idea behind our Explain / Apply / Sustain Mastery Gates. Somebody on that team clearly knows a growth number alone doesn’t prove learning happened. Which is exactly why we’d ask you to ask the same thing of any tool, including ours:
Before you trust a growth claim, ask:
- Does the tool require unaided performance before advancing — or just more practice?
- Can it show you why a student got something wrong?
- Does it check whether the skill is still there weeks later — not just today?
If a tool can’t answer those three, the growth number on its landing page is measuring something, but it might not be measuring learning.
Want a deeper look at how Mastery Gates work in practice? That’s Module 2, Lesson 4 of the Educator Track.
Leave a Reply