Signal

Platform announcement monitoring and industry news monitoring confirmed two developments this week that land on the same pressure point. First, OpenAI launched persistent always-on agents at DevDay that proactively surface answers, documents, and recommendations before a user runs a search. Second, industry news monitoring documented that Wikipedia's page views dropped 8% year-over-year as AI systems return Wikipedia-derived answers without sending users to the source. The same substitution dynamic applies to every information layer buyers use before calling a firm. Boutique advisory and consulting firms that built visibility through directories, articles, and referral-adjacent content are watching that traffic get absorbed upstream — and most have not measured it yet.

Thesis

You built your discovery layer on the assumption that buyers who needed what you do would eventually find their way to a conversation. Someone referred them, they searched, they found an article you wrote, they landed on a directory that listed you. That sequence worked because information lived at destinations — pages, profiles, sources people visited. That architecture is being bypassed. The firms being surfaced in AI-generated answers right now are not winning because they have better relationships. They are winning because their positioning, their language, and their documented expertise made it into the training and citation layer before the shift accelerated. That 8% traffic drop documented this week is not a Wikipedia problem. It is the first measurable evidence of what happens when answers replace visits. The founders who audit their own discoverability this weekend will know exactly where they stand. The ones who wait will find out in a pipeline review three months from now.