Skip to main content

Build vs. Borrow Prompt

Evaluating whether to use existing solutions

Overview​

One of the most expensive mistakes in software engineering is building "scratch" implementations of solved problems. This prompt helps you leverage AI to perform a "Build vs. Borrow" analysis, ensuring you only build what actually differentiates your product.

Key insight: Asking an AI to "survey the landscape" is often more effective than manual Google/npm searches because models have ingested thousands of comparison articles and documentation sites.

The Prompt Template​

Copy and paste this into your AI assistant (Claude, Gemini, ChatGPT):

I need to implement [SPECIFIC FEATURE/PROBLEM] for a [PROJECT CONTEXT, e.g., React Web App, Python CLI].

Please act as a Senior Technical Architect and help me conduct a "Build vs. Borrow" analysis.

### 1. The "Borrow" Landscape
Identify the industry-standard libraries or services for this problem in the [LANGUAGE/ECOSYSTEM] ecosystem.
Select the top 3 candidates based on **community adoption (downloads/stars)**, **recent maintenance activity**, and **relevance to my project constraints**.
For these candidates, please analyze:
- **Maturity**: Community usage, maintenance status.
- **Cost**: Bundle size, performance overhead, or financial cost (if SaaS).
- **Fit**: How well it matches my specific needs.

### 2. The "Build" Analysis
If I were to build this from scratch:
- What is the "Iceberg" complexity? (Hidden edge cases, security risks, maintenance burden).
- What is the estimated time to build a production-grade version (not just a prototype)?
- Is this problem unique to my domain, or is it a generic utility?

### 3. Strategic Recommendation
Based on the above, what do you recommend?
- **Borrow**: If high-quality solutions exist and this is not a core differentiator.
- **Build**: Only if existing solutions are critically flawed for my use case or this is core IP.

Please format the output as a decision matrix or pros/cons list.

When to Use​

  • New Feature Request: When asked to add a calendar, auth, rich text editor, etc.
  • Refactoring: When replacing a buggy custom implementation.
  • Architecture Planning: When defining the stack for a new project.

Example Scenario​

Goal: Adding drag-and-drop to a React app.

Filled-in Prompt:

I need to implement drag-and-drop reordering for a list of cards for a React 18 SPA using TypeScript.

Please act as a Senior Technical Architect and help me conduct a "Build vs. Borrow" analysis.

1. The "Borrow" Landscape​

Identify the industry-standard libraries or services for this problem in the React ecosystem. Select the top 3 candidates based on community adoption (downloads/stars), recent maintenance activity, and relevance to my project constraints. For these candidates, please analyze:

  • Maturity: Community usage, maintenance status.
  • Cost: Bundle size, performance overhead, or financial cost (if SaaS).
  • Fit: How well it matches my specific needs.

2. The "Build" Analysis​

If I were to build this from scratch:

  • What is the "Iceberg" complexity? (Hidden edge cases, security risks, maintenance burden).
  • What is the estimated time to build a production-grade version (not just a prototype)?
  • Is this problem unique to my domain, or is it a generic utility?

... [rest of prompt]

Likely Outcome: The AI will recommend dnd-kit or react-beautiful-dnd, explaining that building accessible, mobile-friendly drag-and-drop from scratch is months of work (the "iceberg").

Tips for Better Results​

  • Plan Mode: If your AI tool supports a "Plan" or "Reasoning" mode, use it. This encourages the model to search its knowledge base more thoroughly before answering.
  • Web Search: Ensure the AI has access to the web if you need the absolute latest library statistics (e.g., "search npm for latest trends").
  • Ask for Prototypes: After the analysis, ask: "Create a minimal prototype using your recommended library to prove it works."