Tao Te Ching Learning AI v0.1.5 — From NotebookLM Study to Limited Beta Preparation

June 9, 2026

TaoTeChingLearningAINotebookLMClaudeCodeLocalMVPLimitedBetaAICollaboration

Summary

I started by uploading Tao Te Ching study materials to NotebookLM. Reading and summarizing documents alone wasn’t enough to verify actual comprehension, so I built a Q&A-based local web learning tool in collaboration with Claude Code. From v0.1 through v0.1.5, the project expanded to include multi-user separation, privacy notice, record deletion, account deletion flow, original text reference panel, and a character dictionary covering chapters 1–10. The current state is not a public service — it is in the pre-public, owner validation and limited-beta rehearsal preparation stage.

The core of this record is not Tao Te Ching interpretation. It is the operational process of systematizing a personal learning project through AI collaboration from local MVP to limited beta readiness.

Background: Why Build a Tao Te Ching Learning Tool

The Tao Te Ching is not easy to approach casually. Classical Chinese source text, multiple translations, scholarly annotations, and modern commentaries all coexist without a clear starting point. NotebookLM became the first step — uploading the source text, translations, and lecture materials to use as an AI-assisted study hub.

Reading AI-generated summaries felt sufficient at first. But over time, one gap became clear: there was no way to confirm whether I actually understood the material or just found it convincing. Comprehension and the ability to express something in your own words are different things.

That gap drove a pivot — from “a tool for reading documents” to “a tool for answering questions and checking understanding.”

From NotebookLM to a Local MVP

NotebookLM continued as the material management and AI query layer. It handled source text search and summary requests throughout the project.

What NotebookLM alone couldn’t provide:

A way to write and track my own answers
Structured progress tracking by chapter and topic
A shareable structure for use with family or close acquaintances later

Those gaps prompted building a local web learning tool.

Role Separation

Roles were clearly defined from the start.

Operator: Set learning purpose, user flow, validation criteria, scope of access, and operational risk. Decided what to build, for whom, and how far to open it.
Claude Code: Designed the technical architecture, wrote all implementation, resolved environment issues, and expanded features.
NotebookLM: Served as the source text, annotation, and commentary reference layer throughout.
Meta-Chulbuji: Handled direction alignment and asset-documentation judgment.

The operator did not write the code. The operator defined the purpose and decision criteria; Claude Code handled implementation.

Technical Stack

Frontend: React + Vite
Backend: FastAPI (Python)
Database: SQLite
Auth: JWT + PIN-based user separation
Runtime: Local development server (frontend + backend running in parallel)

No public deployment server. The application runs locally only.

v0.1 to v0.1.5: Step-by-Step Expansion

v0.1 — Basic Q&A Structure

Chapters 1–5. A minimal structure presenting per-chapter questions and accepting typed answers. Single-user, local record storage.

v0.1.1 — Learning History and Progress Tracking

Added answer record storage and per-chapter progress display. Previous answers became reviewable.

v0.1.2 — Chapters 1–10 Expansion + Original Text Reference Panel

Extended study scope to chapters 1–10. Added a side panel showing classical Chinese characters and translation while answering questions — intended for self-checking against the source while responding.

v0.1.3 — Character Dictionary Expansion

Added brief dictionary entries for key classical Chinese terms in chapters 1–10. These entries are explicitly labeled as beginner-level learning aids, not academically verified interpretations. The following items are marked as requiring further review rather than stated definitively:

Chapter 2: 有無相生 (yǒu wú xiāng shēng) and 無為 (wúwéi) — personal study understanding; interpretations vary among scholars
Chapter 5: 芻狗 (chú gǒu) and 橐籥 (tuó yuè) — descriptions kept minimal to reduce potential misreading
Chapter 8: 上善若水 (shàng shàn ruò shuǐ) — working understanding compiled by the operator, pending review
Chapter 10: 玄德 (xuán dé), 抱一 (bào yī), 玄覽 (xuán lǎn) — limited to beginner-level explanations, not stated as definitive

v0.1.4 — Multi-User Separation and Beta Access Code

Moved from single-user to multi-user architecture. Added a beta access code entry step for participant verification. Each user’s learning records are stored separately.

v0.1.5 — Full Limited Beta Preparation Layer

Added the following:

PIN-based access restriction (including fix for browser autofill override issue)
Privacy notice screen
Record deletion (users can delete their own learning history)
Account deletion / withdrawal flow
Improved JWT-based session management

Runtime Issues Resolved

Port conflict: A prior backend process left behind from a previous session caused port collision at startup. Port status check and process termination were added to the startup routine.

Registration form input bug: Under certain conditions, form input values were being reset during the user registration flow. Fixed by correcting the form state management logic.

PIN autofill issue: The browser’s autocomplete feature was populating the PIN field with previously saved values. Resolved by setting the appropriate autocomplete attribute.

Why It Isn’t Public

Tao Te Ching Learning AI v0.1.5 has not yet gone through the following stages:

Owner validation: The operator has not yet used the tool through a full study session to verify the intended learning flow.
Limited beta rehearsal: The full cycle of external access setup and user flow has not been run end-to-end.
Interpretation review: The beginner-level character dictionary entries need additional review to confirm they don’t create misunderstanding.

Releasing publicly without completing these steps is not appropriate.

Why This Counts as a June Systematization Case

This was not simply “building another app.”

It started from a personal learning purpose, used NotebookLM to organize materials, built a local MVP, expanded it incrementally, and carried it through to limited beta preparation. Throughout, the operator owned judgment, Claude Code owned implementation, and NotebookLM provided the reference layer.

Setting a harness, separating roles, validating incrementally, and recording results as operational assets — this is a concrete example of the June operating theme: systematization, harness-based operation, and AI collaboration.