We don't read your data

What we hear“The vendor's people will be looking at our numbers.”

Vaulty is built and maintained on the shape of your data, never on its content: tables, columns, counts and dates, not names, products or figures. It is a written rule of how we work, enforced by a guard in our tooling.

To make Vaulty understand a company, we need to know how its data is shaped: which tables exist, what the columns are called, how full they are, what period they cover. We do not need to know what is in them. So that is the line we work on. Our engineers, and the AI tools we build Vaulty with, see structure and counts. They do not see a customer's name, a product, a price, a note someone wrote.

How the rule is kept

This is not a preference; it is written into the way we work and repeated to every tool and every person on the project. Our development tooling carries a guard that blocks queries returning content and allows only counts, sums, minimums and maximums. When a repair genuinely needs a value, we stop and ask you for it rather than looking. When something cannot be verified without seeing content, we tell you it cannot be verified.

What this does not mean

It does not mean nobody can reach the node. The node is software, and it needs to be installed, updated and repaired. Named people on our side hold a key to it for that purpose, and that access is for running the software, not for reading what it holds. If you prefer, your IT team can run the node and we guide them.

Last updated: 2026-11-01

The other promises

Your data stays in your building“I don't want our numbers on someone else's cloud.” Nothing you give us trains anyone's model“Will our data make the AI smarter for our competitors?” You decide what Vaulty sees“It will go through everything we have: recipes, contracts, mail.” A breach at Vaulty can't leak your database“If they get into you, they've gotten into us.” Your data is never sold, shared or reused“What do you do with our data on the side?”