Ship-With-AI Evaluate Side Project
Overview
Turn a rough side project idea into a structured build/no-build decision. Optimize for killing weak ideas early, sharpening promising ones, and keeping the MVP constrained to one measurable value proof.
Input
Require one sentence describing the idea. If missing, ask for it before running the evaluation.
Optional context to use if provided: target user, pricing model, distribution channel, constraints, existing competitors, available build time.
Workflow
Run all three checks in order. Be direct; do not make weak ideas sound better than they are.
1. Reddit Problem Mining
Generate 10 messy, emotionally charged, first-person vent phrases a real user might type when frustrated by the problem.
Rules:
- Use informal, specific, late-night language.
- Avoid marketing polish.
- Tie phrases to the proposed target user.
- Suggest 3-5 relevant subreddits or communities.
- List recurring themes to look for in posts and comments.
- Do not fabricate actual post URLs. If the user asks for real URLs, browse/search and label findings as sourced.
2. Kaufman 10-Factor Score
Score each factor from 1 to 10 and give 1-2 sentences explaining the score:
| Factor | What to judge |
|---|---|
| Urgency | Do users actively seek relief now? |
| Market Size | Are there enough reachable buyers? |
| Pricing Potential | Will users pay enough to matter? |
| Customer Acquisition Cost | Can users be reached cheaply? |
| Cost of Value Delivery | Is delivery cheap and scalable? |
| Uniqueness | Is the idea meaningfully differentiated? |
| Speed to Market | Can a useful version launch quickly? |
| Upfront Investment | Can it start with low capital/time? |
| Upsell Potential | Can value expand after first use? |
| Evergreen Potential | Will the problem remain relevant? |
Then total the score out of 100.
Interpretation:
- 80-100: strong candidate; validate fast.
- 60-79: plausible; fix weak factors before building.
- 40-59: risky; only build if learning value is high.
- Below 40: do not spend a weekend on it.
3. PASTA MVP Test
Define a one-feature MVP:
- Job Story:
When [situation], I want to [action], so I can [outcome]. - Success metric: measurable and time-bound.
- Not-Doing list: exactly what is excluded from the MVP.
- Minimal scope: one user, one action, one result, one metric.
- PASTA Skeptic: ask 5 blunt first-time-user questions about value, speed, clarity, trust, and willingness to use/pay.
- Verdict:
PASSorFAILwith one-line rationale.
Output Format
# Side Project Evaluation: [Idea]
## Verdict
- Decision: PASS/FAIL
- Score: [N]/100
- One-line rationale: [...]
## 1. Reddit Problem Mining
### Vent phrases
1. [...]
### Communities
- r/[...]: [...]
### Themes to look for
- [...]
## 2. Kaufman 10-Factor Score
| Factor | Score | Rationale |
| ------- | ----: | --------- |
| Urgency | N | ... |
**Total:** [N]/100
**Summary:** [...]
## 3. PASTA MVP Test
**Job Story:** When [...], I want to [...], so I can [...].
**Success Metric:** [...]
**Not-Doing List:**
- [...]
**PASTA Skeptic Questions:**
1. [...]
**MVP Verdict:** PASS/FAIL - [...]
## Next Action
[One concrete next step: research, interview, landing page, smoke test, or kill.]
Quality Bar
- Prefer a clear
FAILover a polite maybe. - Call out assumptions explicitly.
- Separate evidence from guesses.
- Keep scope small enough to test in one week.
- If two ideas are provided, run the full evaluation for each and recommend the higher-scoring idea.