By August 2026 Zen Aegis had an automated engine that scored how a website shows up in search and in AI answers: 53 checks in four weighted categories, with the evidence for every result stored next to it. It was fast and it was consistent. The open question was whether it was right.
Robby Prochnow did not answer that with a demo. Common Ground already had an expert audit method, done by hand. On August 10 he ran the engine against that hand audit across ten sites and sixteen check types, and then went through every place the two disagreed.
They agreed 92.8 percent of the time with no factual conflicts. Each of the eleven differences had a reason: three were grading scale, one site had changed between the two passes, and seven were subpages the engine had read and the manual audit had not.
Only then did the engine go to work on Common Ground's own sites, where it caught drift nobody had seen: a robots file pointing at a sitemap on a dead domain, a meta description that had slipped back to a placeholder, and a placeholder analytics ID. The point of the hand audit was not to beat the expert. It was to know exactly where the machine and the expert part ways before anyone relies on the score.