Developer: 90% of AI agents are gimmicks; shares 3 workflows that actually made clients money
Affectionate-Ask7235 · reddit · 2026-08-16
A developer who built 40+ AI workflows for clients this year argues that most 'autonomous agents' hyped on YouTube don't work in real business. He cites issues like infinite loops, API cost burn, and context window limits. He shares three profitable workflows: speed-to-lead with n8n and Claude (doubled close rates), competitor price monitoring with Firecrawl and Sonnet (low cost), and human-in-the-loop draft generation with Slack approval. His stack: self-hosted n8n on a $10 VPS, Claude Sonnet, Firecrawl, and webhooks. He advises against fully autonomous agents, recommending strict state machines with AI handling single tasks.
More from coding & agent
- Using AI agents to simplify Apple Search Ads CLI workflows — rudrank · 2026-08-16
- Vibe coding risks: Unreadable codebases defer technical debt to production — pbloemesquire · 2026-08-16
- Open Source Library: Automate Topaz Gigapixel AI processing via Python — tom_doerr · 2026-08-16
- Run Qwen3.8-27B NVFP4 on RTX 5090: vLLM, 256K Context, 147 tok/s — AIFlow_ML · 2026-08-16
- Agents now recover 80% of human NanoGPT speedrun gap, up from 40% a year ago — MinqiJiang · 2026-08-16
- Critique of Vibe-coding: How to Properly Guide Students in AI Era — pbloemesquire · 2026-08-16