01
AI conversation
A little less rehearsal. A lot more real conversation.
PrototypeAI-NATIVE LANGUAGE LEARNING
Learn a language
by actually
living it.
A new way to learn through contextual conversations, adaptive lessons, and AI companions that meet you where you are.
STARTING WITHIDEN日本語
basa polyglotConcept preview
MEET YOUR COMPANIONHikariひかりJapanese, with a little more human warmth.
CONTEXT → REASONING → FEEDBACKFRONTEND PROTOTYPE
Less memorizing.
More meaning.
More meaning.
Language learning should feel like interaction.
Every sentence is a chance to understand more.
01 / THE EXPERIENCE
A conversation.
A lesson. A connection.
Basa Polyglot brings structured learning and contextual conversation together. Starting with Indonesian, English, and Japanese, it is designed to expand across languages.
A LITTLE PRACTICEScripted demo
YOUR NEXT STEP
Want to try another sentence?
Interactive product prototype: responses and exercises are curated. Live AI, voice, accounts, and persistent learner memory are not connected.
02 / LEARNING, CONNECTED
Don’t just get an answer.
Understand it.
Isolated exercises miss the context. Open-ended chat can miss the lesson. We’re exploring a learning experience that brings both together.
01
Say something
Start with a sentence, a question, or a situation.
02
Find the meaning
Consider language, intent, grammar, and useful vocabulary.
03
Make it click
Get an explanation connected to what you were trying to say.
04
Keep it moving
Try a focused next exercise, with room to make mistakes.
A connected learning loop, from context to the next exercise.
03 / MORE WAYS TO LEARN
Your world.
In another language.
From your first sentence to the conversations that matter. One connected product direction, with room to grow.
02
Everyday roleplay
A Tokyo café. A job interview. Your next conversation.
Prototype03
Adaptive lessons
A learning path that responds to what you need next.
Planned04
Writing lab
Go beyond corrections. Understand your choices.
Planned05
Grammar intelligence
The why behind the words, in context.
Prototype06
Vocabulary memory
Bring useful words back before they slip away.
Planned07
Voice & pronunciation
Find your voice, one conversation at a time.
Exploring08
Daily practice
Small sessions, connected to a bigger learning journey.
PlannedPrototype = curated website demonstration · Planned = future capability · Exploring = research direction.
04 / LANGUAGE INTELLIGENCE
Small details.
Deeper understanding.
A sentence is more than a string of words. Basa Polyglot’s intended foundation connects structured multilingual knowledge to explanations that respect context.
- Grammar, with the reasoning behind it
- Vocabulary connected to real situations
- Japanese script, reading, and kanji support
- Learner context that guides the next step
Structured multilingual learning data is part of the planned foundation.
ONE IDEA. THREE LANGUAGES.CONTEXT / PAST
JA
昨日、学校に行きました。
Kinō, gakkō ni ikimashita.EN
I went to school yesterday.
ID
Kemarin, saya pergi ke sekolah.
THE CONNECTION
Japanese and English mark the past in the verb. Indonesian often signals time through context, such as kemarin (yesterday).
05 / UNDER THE SURFACE
Warm on the outside.
Structured underneath.
Aether is the proposed orchestration layer behind Basa Polyglot: connecting learner context, reasoning, and language tools in controlled learning workflows.
Proposed architectureCONTEXT / DESIGN INTENT
Learner context
The companion would organize the learner’s sentence, chosen language, and proficiency into useful context. Learner memory would require clear data controls.
- Language detection
- Intent
- Learner state
Design simulation
This proposed workflow is simulated locally; no model, agent, or retrieval service executes.
TECHNICAL FOUNDATION
Built with frontier
reasoning in mind.
We’re exploring Claude for contextual grammar explanations, multilingual reasoning, and structured tool use. Model integration is planned; this website makes no live AI requests.
- Implemented
- React · TypeScript · Vinext / Next.js APIs
Tailwind CSS · responsive frontend - Experimental design
- Aether orchestration · structured language data
Tool routing · schema validation - Planned / under evaluation
- Claude API · Python / FastAPI · vector search
Learner state · Cloudflare product infrastructure
06 / WHERE WE’RE HEADING
Built for the learner.
Beyond the answer.
The next generation of language learning should understand context, intent, progress, and the person behind every sentence.
- Current phase
Language intelligence foundation
Product website, curated examples, and interaction prototypes.
- Planned
Interactive AI companions
Choose original tutors with different teaching styles for contextual conversations. Deeper customization would follow.
- Planned
Adaptive learning engine
Personalized lessons and progress informed by learner needs.
- Exploring
Voice & pronunciation
Speaking, listening, and expressive character animation, explored together.
- Long-term direction
Expanded languages
Bring the learning experience to more language communities.
A SUSTAINABLE PRODUCT
A future free learning foundation, with optional premium conversations and deeper learning tools. Pricing and paid access will follow a working product.
07 / THE BUILDER
Early stage.
Clear intention.
Bayu Sanjani
FOUNDER / BUILDERBuilding at the intersection of AI systems, language technology, and interactive learning.
View founder portfolio (opens in a new tab)LET’S BUILD THE NEXT CONVERSATION
Learning should feel
like a conversation.
Interested in the future of language learning?
We’d love to hear from you.