video1.0<iframe src="https://www.loom.com/embed/e9ceef793ab149bfad58387e7eb61bce" frameborder="0" width="1920" height="1440" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>14401920Loomhttps://www.loom.com14401920https://cdn.loom.com/sessions/thumbnails/e9ceef793ab149bfad58387e7eb61bce-ded12259f36e058a-full.jpg3682This Loom explains why AI agents like Robo struggle in the wild and how dirty Confluence knowledge undermines trust. Speakers cite Gartner’s prediction that by 2026 around 60% of AI projects will fail because underlying data and knowledge hygiene are insufficient. They describe how AI can blend outdated or duplicate pages into plausible but incorrect answers, leading to business risk, wasted time, and shadow knowledge. To fix it, they recommend starting small with audits of highly trafficked spaces, using governance and permissions, and tracking metrics like adoption, knowledge quality, and business impact. When hallucinations occur, the protocol is to not panic but to track the source, fix the underlying page or structure, and try again.