Automation vs Agentic AI: Most 'Agents' Are Just Scripts, LLM Only Fills the Gaps
forevergeeks · reddit · 2026-09-23
Many workflows marketed as "agentic AI" are just sophisticated scripts. The author illustrates the real distinction with examples: a viral Reddit post about Claude watering plants—something possible since the 90s—got upvoted only because it involved Claude; Nagios has been proactively monitoring infrastructure for over a decade and "might be called AGI by some AI experts today."
The genuine difference is reasoning: traditional automation tells you something is wrong and waits for human troubleshooting, while an LLM plugged into it can investigate the issue, determine a course of action, execute tools, and verify resolution.
Automation follows configured rules and predefined paths (the deterministic foundation); agents determine which path to take when those paths aren't enough. In most agentic workflows, traditional automation can still do most of the work, with the LLM invoked only at the edges.
More from AGI Musings
- Busier than ever but jobs keep vanishing: LLM era demands 10x adaptation speed — dejavucoder · 2026-09-23
- Musk predicts 1 billion humanoids in 10 years, 100 billion in 20, at 10x every 3-4 years — davidpattersonx · 2026-09-23
- Shanahan's role-play paper: describe LLMs with programs, not minds — gerardsans · 2026-09-23
- New frontier model lifted one data firm's ARR from $10M to $300M in 3 months — menhguin · 2026-09-23
- "AI solved the doing; the hard part is now knowing what's worth doing" — ThePeterMick · 2026-09-23
- Gates Foundation-led coalition of 60 orgs aims to bring AI to 3.4B speakers of underrepresented languages — ChinasaTOkolo · 2026-09-23