How-to

How to Turn a Whole AI Conversation Into a Structured Project With One Prompt

The good stuff is stuck in the scroll

A long conversation with Claude can end up genuinely useful — a workflow you built, examples you worked through, decisions you landed on. All of it is real, and all of it is stuck in a scroll you'd have to re-read to find again. Nothing about a chat log makes it easy to hand to someone else, or even to yourself a week later.

Turning that same conversation into something structured used to mean copying pieces out by hand, page by page. It doesn't anymore.

What one prompt actually does

With xTiles connected to Claude, asking it to build a project out of the current conversation is enough — something as plain as "create a project in xTiles based on what we've talked about." Claude reaches for a built-in project-builder skill that's part of the MCP connection itself, so there's nothing separate to install or download first.

Before it builds anything, it asks what kind of project you actually want — a simple personal space or something more detailed — rather than guessing and handing back a structure that doesn't fit.

What you get: a five-page project, not a wall of text

In one run built from a conversation about xTiles and Claude, the result was a five-page project: Home, Daily Routines, MCP Capabilities, Use Cases, and a Log. Each page holds a slice of the conversation in its own place instead of everything sitting in one undifferentiated block — a home overview, a page for recurring workflows, a page for what the connection can do, a page for concrete examples, and a running log.

It builds and edits — it does not delete

Creating a project this way asks for permission the same way any other content-creation request does. Claude can modify existing content in xTiles when explicitly asked to — renaming a task, updating a tile — but it does not delete anything on its own. Building a new project from a conversation adds to your workspace; it doesn't touch what's already there unless you say so.

Where it lands when it's done

Once built, the project shows up as a direct link you can open right away — no separate step to go find it inside xTiles first. From there it's a normal project: shareable, editable, and still there next week, unlike the conversation it came from.

Frequently asked questions

Do I need to install anything to turn a conversation into a project?
No. The project-builder skill is part of the xTiles MCP connection itself, so nothing separate needs to be downloaded or installed first.
What does the resulting project actually look like?
It's a multi-page structure rather than one long document — in one example, five pages: Home, Daily Routines, MCP Capabilities, Use Cases, and a Log, each holding a different slice of the conversation.
Does Claude just guess what kind of project to build?
No. It asks what kind of project you want first — for example, a simple personal space versus something more detailed — before creating anything.
Can this delete or overwrite content already in my xTiles workspace?
No. Claude can modify existing content when you explicitly ask it to, but it does not delete anything on its own. Building a new project adds to your workspace rather than touching what's already there.
What exact prompt do I need to use?
Something as simple as asking Claude to create a project in xTiles based on the current conversation is enough to trigger the project-builder skill.
How do I find the project once it's created?
Claude returns a direct link to the finished project as soon as it's built, so there's no separate step of searching for it inside xTiles.