Mnememory
Most language apps drill a word until it sticks. A word sticks faster when it arrives already attached to something you can see, and almost nothing is built that way.
The constraintThe method has to be honest about its limits: a word a picture cannot hold gets a usage note instead of a forced image.
Founder, designer and engineer of both the product and the public site.
- 10 languages at launch
- 8 review styles
- 5 kinds of assignment
- Learner and classroom in one schema
- 01A keyword picture and a hook for every word
- 02Grammatical gender carried in the image on one fixed rule
- 03A forgetting-curve scheduler held on the server
- 04Eight review styles and a memory palace builder
- 05Classes, five kinds of assignment and a gradebook
- 06The public site around it
A language learning company built on the way memory actually works. Every entry gets a keyword that sounds like it in your own language and a hook you can see, so the word arrives already attached to something.
Gender is carried in the picture on one fixed rule rather than in a table to be memorised: masculine on fire, feminine underwater, neuter frozen. Underneath sits a modern forgetting-curve scheduler that decides when each word comes back, held on the server so the schedule is the same wherever you study.
Built learner-first, with the classroom in the schema from day one, and a per-word view of exactly what a class is losing rather than a single average.