Example Evaluations
Four startup ideas evaluated by IdeaCheck, showing what a PASS, PROMISING, and PASS ON verdict looks like in practice.
"A SaaS tool that automatically generates unit tests for Python code using an LLM, integrated as a VS Code extension."
Strong developer tooling signal on Hacker News. Several threads discuss the pain of low test coverage and the appeal of AI-assisted testing. Distribution via the VS Code marketplace is a proven channel. Risk: LLM hallucinations in test generation could erode trust quickly — correctness is table stakes for this market.
"An app that lets local restaurants create a loyalty stamp card without any hardware — customers scan a QR code to collect stamps."
Multiple HN threads show this category is crowded (Stamp Me, Loopy Loyalty, Square Loyalty) and restaurants churn from these products quickly. Distribution is brutal — each restaurant is a separate sales call. Community consensus: the TAM is real but the economics don't work without scale.
"A B2B API that converts any PDF invoice into structured JSON for accounting software, priced per page."
High-confidence signal. Invoice parsing is a well-established pain with clear enterprise willingness-to-pay. HN discussions reference this as a solved problem that companies keep building in-house — classic build-vs-buy opportunity. LLM accuracy has crossed the threshold where this is now commercially viable. Timing is strong.
"A mindfulness app for remote workers with guided breathing, focus timers, and a weekly wellbeing score."
Extremely crowded (Calm, Headspace, Insight Timer). HN is skeptical of consumer wellness apps — the community questions retention and whether digital tools meaningfully improve wellbeing. No clear distribution edge. Would need a very specific niche (e.g., mindfulness for air traffic controllers) to stand out.