"12 Agents and a Giant System Prompt" Is Just a Workflow Engine, Argues Dev
sull · x · 2026-10-04
- @abishekv argues that the typical "agent architecture"—a dozen agents, a giant system prompt, and a complex orchestration graph—is really just a rebuilt workflow engine with renamed boxes.
- His core claim: agents should be the opposite—one agent with enough context to reason through a problem, act, observe results, and learn how you operate over time.
- Litmus test: if you're still encoding "if this → call that agent" branching logic, you're writing workflows, not agents. Agents are reasoning + action systems, not workflow replacements.
- sull forwarded the post in agreement.
More from coding & agent
- 8 Emerging Standards for AI Agents You Should Know — bibryam · 2026-10-04
- Developer Guesses Open Coding Models Are Less Than Six Months Behind Frontier — heyneighbor · 2026-10-04
- Harness vs. Loop Engineering: The Two Distinct Problems in Building Reliable AI Agents — goyalshaliniuk · 2026-10-04
- ffmpeg-skill turns coding agents into video editors: 42 local FFmpeg tools, auto-verification — udmrzn · 2026-10-04
- Strata calibrate nearly tripled decode speed: 256K context on a 16GB GPU — MoonsvnLyn · 2026-10-04
- One prompt freed 39.7 GB: Claude Code + ccmd MCP safely cleans dev caches — julsimon · 2026-10-04