Explaining to business people why building software is still hard

📝 Discussion Summary (Click to expand)

Three Prevalent Themes

Expectation vs. reality of AI‑assisted coding – Many commenters note that while tools like Claude can generate code quickly, turning that output into production‑ready software still requires substantial human effort (wiring to backends, debugging, feature thinking).

“My job is to wire to our backend data, and a lot of these wiring require me to be in there and actually think about the features. These take time, and I just haven't figure out a way to speed this process up with Claude.” — miranaproarrow
Software is never truly finished; design for continual change – A recurring insight is that software work is never “done”; teams must build with flexibility and anticipate future changes.

“Never really finish building it” is the key insight. The problem with software is that it is never done. There is always another feature you could have… — etothepii
Misalignment of perceived difficulty/ease vs. actual effort – Observations repeatedly show that what business stakeholders view as easy or hard often contradicts the true complexity, leading to mis‑estimated timelines.

“If a business person thinks a change or new program is very easy to do, it is really a very hard project. If a business person thinks the change or new program us hard to do, usually it is a trivial project.” — jmclnx

🚀 Project Ideas

Generating project ideas…

Automatically turns AI-generated frontend prototypes (e.g., from Claude Design) into production-ready full‑stack code by generating backend API clients, data models, deployment configs, and CI pipelines.
Core value proposition: cuts the manual wiring time described by HN users, letting engineers focus on feature logic rather than boilerplate integration.

Key	Value
Target Audience	Full‑stack developers and teams using AI‑assisted UI tools
Core Feature	Parse AI‑generated UI components, infer required endpoints, generate typed SDKs, OpenAPI spec, Docker/K8s manifests, and GitHub Actions workflow
Tech Stack	TypeScript, AST parsers (typescript-eslint), OpenAPI Generator, Terraform CDK, GitHub Actions
Difficulty	Medium
Monetization	Revenue-ready: SaaS subscription ($15/dev/month) with free tier for hobby projects

HN commenter miranaproarrow said: “My manager who vibe coded our entire webapp in claude design… I just haven’t figure out a way to speed this process up with Claude.” This tool directly addresses that wiring bottleneck.
Enables discussion around trust in AI-generated code by providing verifiable, production‑grade artifacts that can be reviewed and audited.

Static analysis service that scores AI‑generated code on security, performance, error handling, test coverage, and style, delivering a checklist of fixes needed for production.
Core value proposition: gives engineering leads an objective “readiness” metric to answer the “why isn’t this production ready?” frustration voiced on HN.

Key	Value
Target Audience	Engineering managers, tech leads, and developers using vibe‑coding assistants
Core Feature	Integrated CI step that runs security linters (Bandit, ESLint‑security), performance detectors, and generates a readiness score with actionable remediation tickets
Tech Stack	Python (Flask backend), Bandit, SonarQube APIs, ESLint plugins, GitHub App, PostgreSQL
Difficulty	Medium
Monetization	Hobby (open‑source core, optional hosted premium checks)

Reflects etothepii’s observation: “If it's easy and the business thinks it's easy it gets done. If it's hard and the business thinks it's hard it doesn’t.” The checklist makes the hidden complexity visible, aligning perception with reality.
Sparks practical utility by turning vague “not ready” feelings into concrete, trackable tasks that can be prioritized in sprints.

Collaborative tool that records the natural‑language prompts given to AI coding assistants, converts them into versioned feature specifications, acceptance criteria, and mock data, bridging product and engineering communication.
Core value proposition: reduces the “never really finish building it” friction by providing a living spec that evolves with the AI‑generated code, making scope clear and manageable.

Key	Value
Target Audience	Product managers, designers, and engineers working with AI‑augmented development
Core Feature	Prompt capture engine, auto‑generation of user stories & acceptance criteria, link to generated code diffs, change log, and real‑time collaboration via WebSockets
Tech Stack	React (frontend), Node.js/Express (backend), MongoDB, Socket.io, GitHub webhook integration
Difficulty	Low-Medium
Monetization	Revenue-ready: per‑seat pricing ($12/user/month) with free trial

Addresses randusername’s point: “once you know software is never done you have to design for constant change.” VibeSpec makes that change explicit and traceable.
Provides a discussion focal point for HN threads about AI‑assisted development, offering a concrete artifact that teams can review and iterate on.