Moving out the loop: durable agents, cron triggers and real near-loop systems
IndyDevDan · youtube · 2026-10-26
IndyDevDan argues engineers should move from human-in-the-loop to proactive durable agents, since frontier models can now handle most terminal-based coding work.
Key points:
- Bridge path "Prompt, Click, Cron": replace human triggers with webhooks, cron jobs, log streams, and Jev classifiers so agents start work on their own.
- Durable agents in sandboxes: dedicated compute, persistent memory, multi-client access (browser/terminal/phone), and crash recovery.
- Three live systems: an on-call incident-response agent (log streaming, 4xx/5xx classification, root-cause analysis, one-click deploy); an always-on news agent running 24/7 for over a week, scoring hype, prompt injection and benchmarks; a UI replicator that hill-climbs toward visual parity via recursive screenshots.
- Skill stack: context engineering, model tokenomics, prompt engineering, and custom harness engineering on top of software fundamentals.
More from coding & agent
- Dev builds open-source Excalidraw MCP after AI-generated diagrams kept overlapping text and arrows — pikahikmag · 2026-10-28
- DeepSeek vs Haiku pricing debunk: accounting for peak rates and token use, 1017 is 2-2.5x cheaper — teortaxesTex · 2026-10-28
- Viral screenshot shows an agent spawning 32 parallel workspaces in ~30 seconds — cephaloform · 2026-10-28
- Grok Bot 0.78.1 hides unreleased per-bot logins with manager-only team sharing — faraz0x · 2026-10-28
- Uber Built a Legal Redlining Agent for Its Lawyers—Here's What It Learned — AxSaucedo · 2026-10-28
- GraphRAG's hub-node trap: same two-hop Cypher costs 71 DB hits normally, 60,013 on a hub — JeremyCMorgan · 2026-10-28