The crack
Selling to enterprises means answering the same security and compliance questions over and over. A buyer sends a spreadsheet with two hundred rows. Someone on the vendor side fills it in, often by copying from an old one. Weeks pass. The deal waits.
The "trust center" that was supposed to fix this is usually a static page with a few logos and a PDF that was out of date the day it shipped. I see this constantly working alongside GRC teams. It is friction everyone has accepted as the price of doing business.
Why it persists
The source of truth lives in scattered policy documents, audit reports, and people’s heads. There is no easy way to let a buyer ask a direct question and get a direct, sourced answer. So the spreadsheet ping-pong continues because it is the only interface anyone built.
The fix on Cloudflare
A living trust center treats the policies and audit evidence as a corpus an assistant can answer from. Built on Cloudflare, that means the documents in object storage, an index over them, and a model that answers a buyer’s question using only those sources, with citations, and a clear "I do not have that" when the corpus does not cover it.
The point is not a chatbot for its own sake. The point is that the answer is grounded in real evidence and is current, so a buyer can self serve the easy questions and a human only handles the genuinely novel ones.
How I am building the demo
I put together a synthetic compliance corpus for a fictional company, Northwind Health: a sample CAIQ, an AI questionnaire, a security whitepaper, and a subprocessor list. That corpus is the ground truth the assistant must answer from.
The grounded assistant runs on Cloudflare’s managed retrieval stack over that corpus. This one is staged behind a small paid tier, so it is the last of the four to go fully live. The corpus and the contract are done; the wiring follows.