---
title: "My AI Skeptic Friends Are All Nuts"
description: "The essay argues that skepticism about LLMs for software development often reflects outdated experience with chat-based code generation rather than current agent workflows. Modern coding agents inspect repositories, edit files, invoke Unix tools and MCP calls, use Git, and run linters, compilers, tests, and model checkers, turning feedback into iterative repairs. The author says these systems can automate tedious typing, lookup, bookkeeping, test refactoring, and edit-compile-test-debug cycles while developers retain responsibility for curation, judgment, direction, and reviewing merged code. Hallucinations are presented as substantially mitigated by tool-based validation, although the author notes that agents remain unsuitable for some trusted operations and that effectiveness depends on language, tooling, prompts, indices, and guardrails. The conclusion is that AI-assisted programming is already materially changing developer productivity, despite its noise and unresolved limitations."
---

# My AI Skeptic Friends Are All Nuts

[Fly.io](https://yomu.fyi/company/fly-io) · Jun 2, 2025

**Type:** Explainer

## Summary

The essay argues that skepticism about LLMs for software development often reflects outdated experience with chat-based code generation rather than current agent workflows. Modern coding agents inspect repositories, edit files, invoke Unix tools and MCP calls, use Git, and run linters, compilers, tests, and model checkers, turning feedback into iterative repairs. The author says these systems can automate tedious typing, lookup, bookkeeping, test refactoring, and edit-compile-test-debug cycles while developers retain responsibility for curation, judgment, direction, and reviewing merged code. Hallucinations are presented as substantially mitigated by tool-based validation, although the author notes that agents remain unsuitable for some trusted operations and that effectiveness depends on language, tooling, prompts, indices, and guardrails. The conclusion is that AI-assisted programming is already materially changing developer productivity, despite its noise and unresolved limitations.

## Context

The author responds to skepticism about LLMs in software development, distinguishing current agent-based workflows from earlier experiences with chat interfaces and pasted code. The essay also addresses concerns about hallucinations, code quality, language support, intellectual property, and developer fatigue with constant AI coverage.

## Approach / What changed

The essay describes coding agents that inspect repositories, edit files, use Unix tools, Git, MCP, linters, compilers, tests, and model checkers, then iterate on feedback. It frames developers as responsible for reviewing and curating the resulting code while using agents for repetitive implementation and debugging work.

## Takeaways

- Coding agents can pull repository and online code into context, modify files directly, run development tools, and iterate through compile and test feedback.
- The author argues that LLM-generated code remains reviewable code; developers are still responsible for understanding and approving what reaches main.
- Agent effectiveness depends heavily on engineering around prompts, indices, tooling, language idioms, and guardrails, not only on model capability.

**Tags:** [AI Agents](https://yomu.fyi/topic/ai-agents), [Generative AI](https://yomu.fyi/topic/generative-ai), [Go](https://yomu.fyi/topic/go)

- Source: [Fly.io](https://fly.io/blog/youre-all-nuts)
- Source URL: https://fly.io/blog/youre-all-nuts
- Ingested by Yomu: 2026-09-02T11:58:27.613Z

[Read original post](https://fly.io/blog/youre-all-nuts)