Power User Details Advanced AI Coding and Research Workflows
TL;DR. A power user shares updated personal workflows for integrating advanced AI models and agents into coding, research, writing, and learning tasks. - The user's coding process leverages frontier models and dynamic workflows for automated project building. - Specific AI tools like ultracode, Codex, and Claude Code are used without custom skills or subagent designs. - The article highlights the practical evolution of AI use, moving beyond basic prompt engineering to complex agent-based systems.
- AI power user details their 2026 workflows for coding, writing, learning, and assistance.
- The core coding workflow involves generating a concept, fleshing it out, creating a tech plan, and then having AI agents build the project.
- The user prioritizes dynamic workflows and advanced models (e.g., ultracode, Codex, Claude Code) over custom skills or manual subagent orchestration.
- This approach allows for 95%+ of code generation in a single 'mega build run'.
- The article serves as an update to how the user leverages AI personally, focusing on practical application.
Sources
- Using AI in 2026 (Coding, Writing, Learning, Assistant-Ing) — blog.sshh.io