IdeaWell

Project ideas from Hacker News discussions.

We charge $10k a week to delete AI-generated code

Original Article

Hacker News Discussion

📝 Discussion Summary (Click to expand)

1. Refactoring AI‑generated “slop” is an emerging business niche

“There is a new kind of task for software engineers these days… 100k lines of AI‑generated spaghetti.” – zie1ony
“Cleaning up after agents with 1M token context is a real business for engineers.” – zie1ony

Companies see huge codebases full of AI‑produced junk and are willing to pay for systematic reduction, even offering price‑per‑deleted‑line models.

2. Pricing and credibility are often questioned

“Your markup on their salaries? … it sounds like they may be at market or below.” – kristianc
“The copy on your website itself kind of reads like LLM slop… it doesn’t inspire confidence.” – neitherboosh

Critics point out that the promised “one‑week, $10 K” deals look more like marketing hype than realistic economics, especially when token costs can balloon quickly.

3. AI is viewed as a productivity tool rather than a replacement for skilled engineers

“We use coding agents too, of course, but as a tool, not as the driving force.” – zie1ony
“AI is a multiplier for experienced, conscientious developers who pay attention. Bad developers can still make bad code with any tool.” – llm_nerd
“The level of ‘slop’ produced by AI is a direct function of skill of the developer and broadness of the prompts.” – adam_arthur

Most agree that AI helps skilled teams move faster, but only when guided by human expertise; raw AI output alone rarely yields maintainable software.

4. The human side: de‑personalized coding and shifting job roles

“It’s de‑personalized… I can tell a junior that a change needs to be refactored without psychological damage.” – supern0va
“We just CMD+Q VS Code and it’s not even in the recents/pin to dock… we just gloss over diffs in the PR.” – thraway3837
“The joy is in seeing the feature come alive, not fighting the computer.” – thraway3837

Engineers increasingly act as overseers and curators rather than primary authors, which changes motivation, review practices, and the perceived value of code quality.

🚀 Project Ideas

Refactor-as-a-Service (Raas)

Summary

  • Solve the problem of paying for AI‑generated code that becomes unmaintainable.
  • Offer a commitment‑based billing model where clients pay only for the amount of code actually removed.

Details

Key Value
Target Audience Engineering managers and CTOs of companies drowning in LLM‑generated codebases
Core Feature AI‑driven refactoring that guarantees a predefined reduction target (e.g., 30% LOC cut) and invoices proportionally
Tech Stack Python + LLMs (Claude Opus, GPT‑4), Bazel for build automation, Docker for isolated environments
Difficulty High
Monetization Revenue-ready: $0.02 per line of code removed (minimum $500 per project)

Notes

  • HN users repeatedly ask “how do we get paid for deleting code?” – this platform makes that explicit and auditable.
  • Built‑in test harness ensures regressions are caught before payment is released, addressing concerns about hidden breakage.
  • Can integrate with existing CI/CD pipelines, letting teams keep their existing workflows.

Code‑Health Inspector

Summary

  • Detect and quantify “slop” in codebases before it turns into technical debt.
  • Provide auto‑generated maintainability scores, test coverage gaps, and concrete remediation tickets.

Details

Key Value
Target Audience DevOps teams, senior engineers, and startups adopting AI‑assisted development
Core Feature Static analysis + LLM‑augmented code review that outputs a “Health Report” with actionable tasks
Tech Stack Rust for fast AST traversal, Node.js dashboard, PostgreSQL for storage, React frontend
Difficulty Medium
Monetization Revenue-ready: $99 per repository per month (tiered discounts for >10 repos)

Notes

  • Directly answers comments like “most of the new Github bloat will just be thrown away” by offering a systematic cleanup path.
  • Includes a built‑in CI gate that blocks PR merges until health thresholds improve, giving engineers a reason to adopt it.
  • Community‑driven rule set allows customization for different domains (e.g., finance, health‑tech).

Prompt‑Driven Architecture Assistant

Summary

  • Guide non‑technical users to craft effective prompts that enforce architectural constraints and generate testable code.
  • Bridge the gap between “vibe coding” and disciplined software design.

Details

Key Value
Target Audience Product managers, junior developers, and small teams without deep engineering expertise
Core Feature Interactive prompt builder that suggests constraints (e.g., “must expose REST endpoint with OpenAPI spec”) and validates generated code against a regression suite
Tech Stack TypeScript/React UI, OpenAI‑compatible API wrapper, Express backend, SQLite for state
Difficulty Low
Monetization Revenue-ready: $29 per user per month (team plans available)

Notes

  • Addresses the frequent HN lament that “most of them haven’t read any more than a few lines of code” by forcing structured reviews before deployment.
  • Generates a “Prompt Ledger” that logs decisions, making audits easier for managers and auditors.
  • Potential to embed into low‑code platforms, expanding market reach.

Slop Extraction & Modularization Toolkit

Summary

  • Convert AI‑generated “slop” into clean, modular components (microservices, libraries) ready for version control.
  • Automate the heavy lifting of extracting bounded contexts and generating adapters.

Details

Key Value
Target Audience Enterprises migrating from monolithic AI‑generated code to maintainable architectures
Core Feature AI‑driven decomposition that produces Docker‑ready services, CI pipelines, and migration scripts
Tech Stack Go for service scaffolding, Terraform for infra, Python LLM pipelines, GitHub Actions for CI
Difficulty High
Monetization Revenue-ready: $0.01 per 10 lines of code migrated (minimum $1,000 per migration project)

Notes

  • Directly tackles the pain point “Cleaning up after agents with 1M token context is a real business for engineers.”
  • Provides a clear migration path, reducing the “ball of mud” fear that stops companies from adopting AI‑generated code.
  • Can be packaged as a SaaS with a UI for selecting reduction targets and exporting migrated repos.

Read Later