There Will Never Be a Wikipedia Page for Your Business

I was telling a friend about TrueSignal last week. His reaction: "So it's like Wikipedia. A trustworthy source of facts about local businesses."

Not exactly, but I liked the analogy. Here's where it holds, and where it breaks.

AI doesn't trust Wikipedia because it's famous. It trusts Wikipedia because the information is inspectable. Claims are sourced. Edits are tracked. Disagreements are visible right on the page. The process itself creates the credibility.

Service businesses have no equivalent. There is no Wikipedia page for an HVAC contractor, a dental practice, a family law firm. There never will be.

English Wikipedia has about 7 million articles, covering all of human knowledge. There are more than 30 million service businesses in the US alone. Nobody is building and maintaining encyclopedia pages for neighborhood plumbers, dentists, electricians, and law firms.

So what does AI actually use when it answers questions about a business? Yext answered that in October 2025. They analyzed 6.8 million AI citations across ChatGPT, Gemini, and Perplexity and found that 86% came from sources businesses already control: their own websites (44%) and their listings (42%).

When AI describes a business, it's mostly repeating what the business published about itself. Marketing copy. Directory listings. Self-reported claims. None of it checked against how the business actually operates.

Better copy doesn't fix this. The problem isn't how businesses describe themselves. It's that description is all AI has to go on.

Jimmy Wales has been warning about exactly this failure mode, most recently in a WSJ interview on AI and misinformation. AI generates the most plausible answer, and the most plausible thing isn't the same as the most true thing. He's seen AI cite books that don't exist, complete with realistic ISBNs. "The errors are plausible," he told a room of news executives in Berlin this May. "You might not think to dig in."

That's the gap we built TrustRecord for.

Every TrustRecord is computed, not written. That's where my friend's analogy breaks. Wikipedia is authored by volunteers citing secondary sources. A TrustRecord isn't authored at all. It's computed from read-only connections to the business's own systems of record: field management software, accounting platforms, state registries.

Instead of saying "we've served thousands of customers," the record computes the actual number. Instead of claiming "most of our customers come back," it calculates the repeat customer rate. Instead of "serving the metro area," it measures where work was actually completed. Instead of "decades of experience," it computes years in operation from the accounting record. Every metric carries its source and an as-of date. Data we verified independently is labeled Verified. Data the business declared is labeled Attested, and says so on the record.

The registry doesn't rank businesses. It doesn't recommend them. It documents them. AI systems, and the people asking them, decide.

That's the stance that made Wikipedia citable in the first place: neutrality plus provenance.

So my friend was mostly right, with one correction. Think of it as the Wikipedia entry for your business — except sourced from your actual operating systems, not volunteer editors.

Marketing tells AI who you say you are. A TrustRecord shows what actually happened.