Lean AI Tool Stack
A lean AI tool stack is an operating discipline for keeping only the few AI tools that materially support daily or weekly work, while treating the rest as specialists, experiments, or retired tools.
Key points
- Nate’s working stack is intentionally narrow: Claude Code, VS Code, and Glaido as daily drivers; Codex (OpenAI), Claude chat, Hermes Agent, Perplexity, and Grok as weekly companions [src-053].
- Specialist tools are not daily work surfaces. Nate uses key.ai, GPT Image 2, Nano Banana / Imagen 3 (Nano Banana 2), HeyGen, ElevenLabs, Apify, OpenRouter, and Claude Design only when a specific task calls for them [src-053].
- Experimental tools stay in the ecosystem without becoming default behaviour. Nate names Gemini / Anti-Gravity, Ollama, and Manifold as tools he watches or tests without building his core day around them [src-053].
- Graduated tools are tools whose useful features have been absorbed into the user’s own system or replaced by better defaults. Nate puts ChatGPT chat, OpenClaw, Cursor, NotebookLM, Poppy AI, Anytype, and Whisper Flow in this bucket [src-053].
- The point is not tool minimalism for its own sake; it is reducing cognitive overhead so new AI announcements do not constantly reset the user’s working system [src-053].
Related entities
Related concepts
- Agent Harness Portability
- AI Tool Adoption Decision Framework
- Needle Moved Per Hour
- Productivity Dip Curve
Source references
- [src-053] Nate Herk — “Overwhelmed By AI? Just Copy My Tech Stack” (2026-05-08)
Recommended next
Keep reading from this thread
From 494 indexed pages and articles.
- Wiki concept Perplexity An AI search and research tool Nate Herk uses as a research layer inside automations and agents, rather than as a primary daily Related by lean
- Wiki concept Grok X's AI assistant. Nate Herk uses it mainly for research over X/Twitter threads, posts, and social-platform-native context. Related by lean
- Insight AI Measurement and Experimentation How to measure AI product impact with evals, adoption metrics, online experiments, guardrails, and cost tracking Readers have engaged with this next