Start With the Problem, Not the Agent: A Better Way to Find Agent Building Ideas
forevergeeks · reddit · 2026-09-20
The author argues that people build agents the way they buy software: demo first, problem later. Starting from "what can this agent do?" leads to an endless loop, since agents can interact with software, retrieve information, make decisions, call APIs, and move data.
His recommended order:
- Define the problem first
- Map the workflow: steps, decisions, systems accessed, repetitive tasks, bottlenecks, handoffs
- Build the agent around that workflow
His core point: any task doable through a computer can likely be partly handled by an agent — the trick isn't figuring out what agents can do, but what they should do.
More from coding & agent
- Dev forks MiniMax's terminal coding agent, maintains separate branch with security updates — jasonkneen · 2026-09-20
- Kevin Rose built an offline Grok Bot that turns years of saved IG videos into a searchable md wiki — minchoi · 2026-09-20
- How do teams validate citations in production RAG? Duplicate-source checking called too narrow — No-Profile-6178 · 2026-09-20
- Dev spent a quarter saying "I'm on it" with nothing to show for it — cto_junior · 2026-09-20
- LangChain's Jev-as-a-Judge: cheap, fast semantic verifiers for agent evals — multiply_matrix · 2026-09-20
- Research agents' hidden failure mode: citations that look independent but share one source — No-Profile-6178 · 2026-09-20