What is Real Problem AI?

Real Problem AI is a curated directory of 100 real problems worth building an AI startup around. Each problem is scored honestly on severity, AI feasibility, market signal, and competition gap. We're for builders looking for what to build, not for ethics debates about AI.

What is the difference between Real Problem AI and AI ethics or AI risks indexes?

Real Problem AI catalogues problems AI can solve — productive, commercial opportunities for builders. AI ethics indexes catalogue problems with AI itself (bias, environmental cost, copyright, consciousness). Different intent, different audience: founders and indie hackers vs researchers and policy folks.

What should I build with AI in 2026?

Look for problems with high severity (it costs people hours, sleep, or revenue every week), high AI feasibility (LLMs, vision or voice are great at this today), strong market signal (Reddit threads + willingness to pay), and a clear competition gap (incumbents miss the same obvious thing). Real Problem AI scores 100 such problems for you on those four axes.

Where do Real Problem AI's startup ideas come from?

From three sources, scanned each cycle: 1) Reddit threads where people describe a real friction in their own words, 2) podcasts and talks of founders young builders trust (Nikhil Kamath, Aman Gupta, Ankur Warikoo, Raj Shamani, Andrej Karpathy, Greg Isenberg), and 3) a continuous 1,000+ source scan covering forums, app store reviews, regulator filings, and field interviews.

Is Real Problem AI free?

Yes. Browsing every problem, opening every score breakdown, raising your hand on a co-founder match, and the founder vault are all free. There is no paywall.

How is each problem scored?

Four axes, each 1-10, weighted: Problem Severity (30%), AI Feasibility (25%), Market Signal (25%), Competition Gap (20%). Weighted average is the Opportunity Score on each card. Only problems clearing 7.0+ make the live list.

Why does evaluating something as simple as 'clarity' still require running a thousand prompts by hand?

AI agent builders doing evals face outputs that aren't binary right-or-wrong, forcing expensive, high-volume manual testing just to get statistically meaningful signal on subjective quality dimensions.

Category: AI / Agents · Trend: LLM · Opportunity score: 6.2 / 10

What is the “Why does evaluating something as simple as 'clarity' still require running a thousand prompts by hand?” problem in 2026?

Who has this problem?

Indie AI builders and small teams trying to eval their own agents/prompts.

Recorded source context

Dataset source note: it's hard to measure outputs that aren't binary right or wrong, and that's where most human intervention is needed... it's quite expensive to do well. I find that for every hypothesis I might have to run a thousand prompts to collect enough data for a conclusion.

This note may summarize the referenced material rather than quote it verbatim. Source label: Ask HN: How are people doing AI evals these days?, Hacker News item 47319587, 17 Mar 2026. (reference).

Existing players in this space

Promptfoo / Braintrust / DeepEval: Handle structured evals well, subjective/dimensional quality still needs heavy human review.
LLM-as-judge: Cheaper than humans but introduces its own noise and bias.

What existing players are missing

An eval workflow purpose-built for subjective, multi-dimensional quality (tone, clarity, helpfulness) that combines cheap LLM-judge triage with targeted, minimal human review instead of all-or-nothing manual scoring.

How Real Problem AI scores this opportunity

Aggregate score: 6.2 / 10. Four-axis rubric:

Problem severity: 7 / 10
AI feasibility today: 6 / 10
Market signal: 6 / 10
Competition gap: 6 / 10

How to build a solution: stack hints

LLM-as-judge scoring pipeline
Active-learning sample selection for human review
Multi-dimensional rubric scoring
Statistical confidence reporting