Hey HN, I’m André, one of the people that is building benchouse. ai, a public leaderboard for analytics agents. To score every agent fairly, we need to give them equivalent semantic context.
Source: [Hacker News](https://github.com/benchouse/semglot)