An impressive demonstration is not the same as a durable opportunity. Evaluation should begin with the person or organization experiencing the problem, then test whether AI can produce a reliable and economical improvement.
Core evaluation framework
| Factor | Question |
|---|---|
| Pain | Is the problem urgent, frequent, expensive, or emotionally difficult? |
| Existing spending | Do users already pay money or spend significant time? |
| Ordinary-user benefit | Does the solution improve access, cost, speed, quality, or control? |
| AI feasibility | Can the task be performed reliably with current or near-term technology? |
| Human oversight | Where must a person review, approve, or accept responsibility? |
| Customer access | Can the builder reach buyers without enormous sales expense? |
| Data | Can the necessary data be obtained legally, ethically, and economically? |
| Integration | How difficult is connection to existing software and operations? |
| Competition | Is the market crowded with interchangeable products? |
| Defensibility | What becomes stronger as customers use the product? |
| Regulation and liability | What happens when the system is wrong? |
| Capital efficiency | Can a useful first version be launched with available resources? |
| Time to revenue | Can paying customers be reached within a reasonable period? |
User-alignment test
- Who is the system economically designed to serve?
- Can the user understand when AI is involved?
- Can decisions be corrected or appealed?
- Is pricing transparent?
- Are privacy and data rights clear?
- Can people with limited technical skills use it?
Capital bands
- Bootstrap: under $250,000
- Small startup: $250,000–$2 million
- Venture scale: $2–$20 million
- Capital intensive: above $20 million