About This Launch
RealRun runs the same everyday tasks on AI models and publishes the scores, the mistakes and how it tested. The site reports 473 test items, 13 models and 9,435 graded answers across five areas: inbox triage, meeting scheduling, receipts, finance and customer support.
Results, a read-only MCP server and a weekly email are free. Tested playbooks are $29 at the founding price ($49 regular), or $99 for all five. A paid lookup API is planned and not live.
Why I Built This
RealRun tests AI models on the everyday jobs people actually hand them, then publishes the scores, the mistakes and the method. The catch: scores reflect our tests on the dates shown (October 2026), not a guarantee, and top models tie on some tests.
Launch Details
CategoryAI / ML
StageLive
ListedOctober 19, 2026