omarsar: Stop automating everything — help your coding agents write better code instead
omarsar0 · x · 2026-11-21
AI writer omarsar shares how his software engineering and research workflow has changed:
- Early mistake: trying to automate everything at once with a pile of subjective skills.
- What works now: spending most of his time helping coding and research agents write better code and explore better ideas — being more involved, not less.
- Core pattern is "loops": automations plus clear, intentional steps he keeps refining, strengthened via verifiers, deterministic steps, context formatting and awareness, memory systems, and evals.
- Takeaway: the job has fundamentally changed — it's not about using LLMs to help us write code, but helping the LLM write code better and faster. This has sharply cut error rates and enabled real scaling with coding agents.
- Outlook: coding agents only took off about a year ago; primitives like agent swarms remain unsolved. He advocates deep domain mastery to prepare for what's next.
More from coding & agent
- Dev launches remote MCP server for live Pinnacle sports odds, queryable from Claude Desktop — Significant-Sun-9735 · 2026-11-21
- The speed limit for AI science is the cost of verification, field by field — professor_ajay · 2026-11-21
- MCP LINE Server lets AI assistants send LINE messages via Messaging API — modelcontextprotocol · 2026-11-21
- icf-mcp serves WHO ICF codes and 11 clinical assessment instruments — modelcontextprotocol · 2026-11-21
- 4 months of RAG-LCC: query-action splitting and table retrieval lessons — HarinezumIgel · 2026-11-21
- A project manager's working review: Pebble Index 01 vs Sandbar Stream AI ring — Scobleizer · 2026-11-21