About Kumi
Built by Learners, For Learners
We're building the Japanese app we wished existed when we started learning. One that doesn't waste your time.
Why Kumi Exists
Most Japanese learning tools break language into disconnected pieces. Grammar lives in one place. Vocabulary in another. Kanji somewhere else entirely. Learners are left to stitch everything together on their own.
That fragmentation makes progress harder than it needs to be.
Kumi exists to unify what other tools keep separate.
Grammar, vocabulary, kanji, and dictionary lookup unified into a single system that reflects how Japanese is actually used.
Instead of forcing learners down preset paths, Kumi adapts to what you already know and builds forward from there. We're not chasing engagement numbers. We want you reading.
Meet the Founders
Building the Tool We Needed
Half Japanese and half American, Yoshimaro grew up without access to his Japanese language or cultural roots. That disconnect led to years of studying Japanese through textbooks, apps, and native material, often feeling that existing tools missed how the language is actually used.
Kumi was built out of that frustration. With a background in computer science, Yoshimaro set out to build a learning system that starts from a learner's current experience and builds from there.
Currently studying at an upper-intermediate level, roughly N3 to N2, and still learning alongside everyone else.
Doylan came into Kumi with no background in Japanese. Rather than starting as a learner, he came at it from the systems side: take the language apart, find the structure, and build the curriculum from that.
Doylan builds tools he actually wants to use. Kumi is that mindset applied to a hard language: approachable without dumbing it down.
As Kumi takes shape, Doylan is learning Japanese alongside its development, and that outside-in view is exactly what keeps the system honest. If something confuses him, it gets fixed.
Professionally Validated
Data You Can Trust
Maria Ando
Linguistic Validator, Grammar Dataset
Every entry in Kumi's grammar dataset is professionally validated. Maria led the linguistic validation and data cleanup behind it, checking every entry against how Japanese is actually used, not just whether it is technically correct.
Having studied Japanese through every JLPT level and passed them all, she pairs that expert eye with a genuine learner’s perspective. She knows firsthand where learners get stuck.
What We Stand For
We believe learning tools should serve learners first, metrics second.
That means no manufactured urgency. No artificial streaks. No gating basic functionality behind layers of plans. Progress should come from understanding, not pressure.
We keep Kumi calm and honest about what learning actually takes. If a feature just keeps you busy, it doesn't ship.
We're not promising disruption. We're building a tool that respects how learning works, and the people doing it.
Come build it with us
Get Started
Be part of building a better way to learn Japanese.