IdeaWell

Project ideas from Hacker News discussions.

Where is all the new, good software?

Original Article

Hacker News Discussion

📝 Discussion Summary (Click to expand)

Theme 1 – AI speedups are often turned into leisure time rather than higher output
Many commenters argue that the productivity gains from AI are being “captured” as extra breaks or personal time, not shipped features.

“Software teams are capturing a lot of the speedup in the form of more leisure or quasi‑leisure time. (To greatly simplify: software teams are offering 2x speedups to their companies. They're then using tools that allow them to work 10x as efficiently, and taking nice long coffee breaks.)” – dude250711

“This might be the biggest factor… an employee who suddenly found a way to multiply their productivity might be keeping that method to themselves and freeing up some time instead.” – pu_pe

Theme 2 – AI‑generated code tends to be buggy, fragile, and lowers overall quality
Several users point out that while AI can produce code quickly, it often creates hard‑to‑maintain output that shifts the burden back onto humans.

“We are required to read and reason about all generated code, but there is a limit how much each person can process in a day.” – kvgr

“AI written software doing the exact thing it was supposed to but breaking apart on any further improvements or changes… the code is often bloated and fragile.” – contravariant

“I perceive a distinct and massive drop in software quality over the past few years.” – sunaurus

Theme 3 – The useful software built with AI stays niche or fails to reach a broad audience
Even when AI helps create good tools, distribution channels are closed, market saturation limits adoption, and many projects remain personal or internal experiments.

“AI coding is great but the distribution channels have been closed for a while. App stores are effectively closed for new apps and old internet is dead.” – cft

“We are in fact seeing software improvement, but the ‘we’ is very limited. Most people don’t adopt new software very often, so it’s not surprising that they aren’t finding new great software now.” – vincent‑uden

“I personally made a bunch of net new, really good software that I use myself… You will not see that software though.” – destiner

🚀 Project Ideas

CodeSentry: AI‑Generated Code Quality Guardian

Summary

  • Continuously scans AI‑produced code for bugs, security flaws, and maintainability issues, auto‑generates test suites and suggests fixes in real time.
  • Core value: reduces human review bandwidth while ensuring AI‑written software meets production‑grade reliability.

Details

Key Value
Target Audience Software teams using LLMs for code generation (frontend, backend, infra)
Core Feature Automated static/dynamic analysis, test generation, security scanning, and auto‑fix PRs for AI‑authored code
Tech Stack Rust (analysis engine), TypeScript (VS Code/GitHub Action), PostgreSQL (findings DB), OpenAPI for CI integration
Difficulty Medium
Monetization Revenue-ready: SaaS tiered pricing ($20/dev/month for basic, $80/dev/month for enterprise with SLA)

Notes

  • HN users complained about “bloated and fragile” AI code and limited human bandwidth to review it (kvgr: “We are required to read and reason about all generated code, but there is limit how much can each person process in a day”). CodeSentry directly addresses that by automating the review loop.
  • Would spark discussion on balancing AI speed with safety, and could be integrated into existing CI pipelines to prevent “move fast and break things” scenarios.

MicroApp Forge: Personalized Micro‑Software Builder

Summary

  • Lets users describe a tiny, single‑purpose tool (e.g., a photo‑crop utility, a custom invoice generator) in natural language; the system outputs a highly optimized native or WASM app with built‑in performance profiling and minimal dependencies.
  • Core value: empowers non‑experts to get reliable, lightweight software tailored to their exact need without the bloat of “vibe‑coded” clones.

Details

Key Value
Target Audience Power users, freelancers, small businesses who need custom micro‑tools but lack dev resources
Core Feature NL‑to‑code pipeline that generates native (Rust/Go) or WASM apps, includes worker‑thread scaffolding, size‑optimization, and one‑click publish
Tech Stack LLM (codegen), Rust/WASM toolchain, Tauri for desktop bundling, Svelte for UI, GitHub Actions for CI
Difficulty High
Monetization Revenue-ready: Subscription ($9/mo per active micro‑app) or marketplace revenue share

Notes

  • Commenters lamented that AI‑generated Photoshop clones are “way too slow in WASM” and “like gargling dog shit” (YuechenLi, sdcfgy). MicroApp Forge would enforce performance best‑ practices (workers, native compilation) from the start.
  • Enables the trend of “small software that is tailored perfectly to the needs of a single user” (ThePhysicist) while keeping distribution simple.

ExplainAI: AI‑Code Comprehension & Knowledge Hub

Summary

  • Turns AI‑generated code bases into living documentation: auto‑generated explanations, dependency graphs, and onboarding guides that update as the code evolves.
  • Core value: lowers the cognitive load for humans to understand and maintain AI‑written software, addressing the “human bandwidth” bottleneck.

Details

Key Value
Target Audience Dev leads, maintainers, and new hires inheriting AI‑generated projects
Core Feature Semantic code summarization, visual call‑graph generation, change‑impact alerts, and Q&A chatbot trained on the repo
Tech Stack Python (ast, graph‑lib), LLM (embedding + chat), Neo4j (graph), React (docs UI), Docker
Difficulty Medium
Monetization Revenue-ready: Per‑repo pricing ($15/mo) with free tier for open‑source

Notes

  • kvgr highlighted the limited human capacity to read/reason about AI output; ExplainAI provides “instant comprehension” so teams can trust AI code without spending hours dissecting it.
  • Would alleviate fears from leadership that they must “take full responsibility over AI output” by providing transparent, continuously updated knowledge artifacts.

Read Later