One-Soup One-Dish: A Japanese Cooking Principle for AI Fatigue

Developer and creator Takuya, writing on his blog Devas Life, draws a parallel between the Japanese cooking principle Ichiju Issai (one soup, one dish) and the overwhelming pace of AI tooling. The core idea: strip away the unnecessary to maintain mental health and creative focus.
The Problem: AI Fatigue
New AI services emerge daily, and major releases from big companies change workflows weekly. Chasing every hype doesn't increase security — it clouds the essential skills developers need. Takuya calls this AI fatigue.
The Solution: Ichiju Issai for Tech
Ichiju Issai is a meal style centered around rice, one soup, and one side dish. Chef Yoshiharu Doi proposed it to free home cooks from the pressure of creating elaborate meals. Takuya applies this to software: treat your core development environment (editor, AI coding agent, or framework) as the rice — the constant that never changes — and allow yourself only one primary tool and one secondary tool. The goal is to create a rhythm where you can return to a comfortable, heart-centered place.
Practical Takeaways
- Decide what NOT to do. Doi says: “By simplifying meals to Ichiju Issai, cooking becomes free of stress.” Your tech stack should follow the same rule.
- Find tools you never get bored with. Just as a parent’s cooking brings comfort through familiarity, your core tools should feel like home.
- Avoid tracking gossip or drama. Social media algorithms try to grab attention; Takuya advises deliberately ignoring them to preserve calm.
- Direction over strategy. Moats can be flexibly changed; the important thing is knowing where you want to go as a developer and artist.
Takuya’s article includes a video version and references Doi’s book 一汁一菜でよいという提案 (The Proposal for One Soup, One Dish).
📖 Read the full source: HN AI Agents
👀 See Also

Running Qwen3.6-35B-A3B with ~190k Context on 8GB VRAM + 32GB RAM – Setup & Benchmarks
A Reddit user shares a working llama.cpp configuration for Qwen3.6-35B-A3B GGUF models on an RTX 4060 (8GB VRAM) + 32GB DDR5, achieving 37-51 tok/s at 192k context using TurboQuant and specific flags.

Building a Custom Hindi Glossary System with Claude: From 76% to 92% Accuracy in 10 Months
A solo dev in Bangalore built a custom glossary system for Claude to improve Hindi domain vocabulary accuracy from 76% to 92%. Example-based terms with context sentences worked best.

OpenClaw installation hurdles on Windows 11 and how to overcome them
A user details three specific obstacles when installing OpenClaw on a fresh Windows 11 machine: PowerShell execution policy, Windows Defender blocking, and missing dependencies like Node.js and Git.

Using Claude to analyze writing patterns for better custom instructions
A Reddit user describes a method for creating more effective custom instructions by having Claude analyze 10 writing samples to identify concrete patterns like punctuation avoidance and analogy sources, rather than relying on subjective tone descriptions.