If you're new to AI-assisted development, it's a fair question to ask before you dive in: once an agent is helping you build on Xperience by Kentico, where does your project's data actually go?
The answer isn't a single connection. A KentiCopilot setup is made up of several components, and they don't all behave the same way. Some see very little and forget it immediately. Others see a lot, by design, because that's what makes them useful. Knowing which is which lets you use KentiCopilot with confidence instead of crossing your fingers.
A few highlights:
- The Management MCP server, the only KentiCopilot component that can reach your project content, runs on your machine and is meant for local development only. It doesn't expose contacts, form submissions, members, or orders.
- The Documentation MCP server never logs or stores your search queries.
- The Content Modeling MCP server validates your draft content model in memory and then discards it.
- The biggest data flow isn't to Kentico at all. It runs between your AI tool and your model provider, under that provider's terms.
The full guide covers all of this in detail. It includes a data flow diagram showing what stays on your machine and what leaves it, plus a plain-language breakdown of what each component can reach, what it records, and what it discards.
One thing to keep in mind: Kentico builds the tools and documents exactly how they behave, but you decide how they're used on your project. That means whether to enable the management API, which tools an agent can call, and which AI tool and model provider you trust with your data. The guide is there to make that an informed choice rather than a guess.
Does your team already have a rule for this that's worth sharing? Let us know in the comments.
