I design AI-native workflows, editorial agents, and knowledge systems.
- AI-assisted editorial systems
- topic-specific research radars
- agent workflows for software and knowledge work
- documentation-first frameworks
- reusable prompts, skills, checklists, and operating guides
A reusable framework for turning fast-moving information streams into curated channels and living knowledge bases. Includes a build guide, a design-pattern catalog, and a real two-channel case study (two live Telegram channels on one shared engine). → github.com/SoulAtelier/ai-native-newsroom
A fork-ready template pack for giving AI agents a clear editorial contract: domain brief, source registry, triage rules, publishing guards, and a knowledge loop. → github.com/SoulAtelier/ai-newsroom-agent-configs
I build through agents, but not by abdicating judgment to them.
idea → architecture → domain rules → agent instructions → implementation → human review → reusable methodology
The interesting part is not only the code. It is the operating system around the code: how the project knows what matters, what to ignore, how to fail safely, and how to remember what it learned.
- AI-native editorial infrastructure
- agent-readable documentation
- knowledge bases that evolve from real workflows
- practical methods for non-developer product builders working with coding agents
Everything is in the repositories pinned to this profile — start there.