Periodic cognitive test that tracks the mental effects of heavy AI usage
Problem
Knowledge workers who use LLMs all day are starting to worry they are losing their ability to think unaided (memory, recall, reasoning), but have no way to measure it. There is no product that lets someone periodically assess their cognitive performance and correlate it with their AI usage — the only advice available is 'try coding without AI for a day'.
Opportunity
A 'cognitive checkup' app: short validated-style periodic assessments (recall, reasoning, focus) plus self-reported AI-usage tracking, producing a personal trendline of cognitive fitness vs. AI reliance. Rides a fast-growing anxiety among heavy AI users and could expand into team/wellbeing plans.
Market analysis
The anxiety is real and accelerating (2025 studies, MIT Media Lab coverage, mainstream press), and no product yet pairs cognitive testing with AI-usage tracking. But the scientific ground is thin: without validated instruments and objective usage data, the trendline risks being astrology for developers.
Market · Heavy LLM users in tech — developers, writers, analysts — riding a fast-growing, media-amplified anxiety wave.
Pricing · Brain-training comparables run $10-15/month (BrainHQ, CogniFit territory); a plausible wedge is $5-8/month personal or $29+ team/wellbeing plans.
Pros
- + Genuine, fast-growing anxiety with strong 2025 research and media tailwind.
- + No existing product combines cognitive tests with AI-usage correlation.
- + Periodic checkups create natural retention loops (weekly/monthly habit).
Cons
- − Self-reported AI usage is unreliable; objective tracking (API tokens, screen time) is hard.
- − Non-validated tests invite the 'pseudoscience' critique that killed brain-training hype.
- − Human Benchmark already offers free tests with score history — the test layer alone is a commodity.
- − Novelty-driven churn: most users will check twice and leave.
Existing / similar tools
- → Human Benchmark ↗
- → Global Mind Tests ↗
- → BrainHQ
- → CogniFit
Source
Hacker News (Ask HN)
The honest version of this product is a tracking journal, not a diagnosis — and positioning it that way is both the ethical move and the defensible one. The non-obvious technical opportunity is objective usage data: local measurement of AI-assisted vs. unassisted activity (IDE telemetry, browser time on AI tools) would beat self-reporting and is something none of the brain-training incumbents can copy quickly. The killer feature would be a periodic “unaided challenge” — the same coding or writing task with AI off — which converts the vague Ask HN advice (“try a day without AI”) into a measurable baseline. Without something like that, the trendline measures practice effects on the tests themselves, a confound that sophisticated users will call out immediately.