What 1.5 Years of Building Agents Taught Me: It's All a While Loop, Tools, and Specs
khant_dev · x · 2026-11-20
Developer parthshr370 published a 9-minute blog distilling 1.5 years of agentic engineering experience:
- It's all a while loop: beneath the hype, every LLM harness/agent reduces to history=[usermessage]; while True: reply=llm.call(...) — agentic trends are "reheated, rehashed versions of the same dish."
- It's also just a tool call: how tools and reflection actually work internally.
- The subagent dilemma: agents ultimately boil down to plain backend engineering.
- How to learn: skip chasing every new model or "this will take my job" release; old material is where the gold is. He suggests asking Claude to diagram the run loop and tool internals with Excalidraw.
He got started in the langchain/camel-ai era and calls this his attempt to condense all the "alpha" into one coherent mental model for catching up with the agentic race.
More from coding & agent
- TwinLab-Bench tests AI scientists in agentic twins of real labs; best model scores just 45.0 — zju-community · 2026-11-20
- Building an AI phone agent: 2018's duct-tape pipeline vs 2026 — JafarNajafov · 2026-11-20
- Open-Source AI Agent Eigin Will Pay Contributors for Specs, Not Bounties — danielrothmann · 2026-11-20
- X creator shares how agents now handle scheduling, research and data analysis — superinvestor_43 · 2026-11-20
- LoRA adapter lifts Qwen3.8-27B to 63% on SWE-Bench Pro, 60% fewer tool calls — ben_burtenshaw · 2026-11-20
- Mubit + Neon: agent middleware cuts frontier-model tokens by 73% — JeremyCMorgan · 2026-11-20