I Think I Accidentally Built an Ecosystem
I realized something while building UluP:
I wasn't building a bunch of unrelated products.
I was slowly building different ways to interact with the same underlying system.
UluP Spaces is at the center.
Then:
Desktop — files as maps
Life — life as a constellation
Spatial — work as space
UluPy — the AI layer
MCP — the bridge between AI and UluP
The interesting part is that these don't have to replace each other.
You can use them independently, but they can also connect.
For example, you can organize files in Desktop, turn them into a project in Spaces, interact with that project through AI, and eventually step inside it through Spatial. (more soon about this one)
That's the direction I'm taking UluP.
It's still early, and a lot of this is experimental.
But for the first time, I feel like I can actually see what UluP is becming.
One ecosystem. Different ways of thinking, working and creating.
Does this architecture make sense to you, or does it feel like I'm trying to build too much?
What would you need to see to know you’re actually building an ecosystem rather than accumulating adjacent products around the same underlying idea?
An ecosystem is a collection of products united not only by a core concept but also by functionalities that link the apps together—and that is precisely what I am building. As shown in the map, Spaces is the flagship product. Desktop and Life connect directly to it, demonstrating that the apps do not operate in isolation but are interconnected; furthermore, all three share a common element: Ulupy, the AI that unites them via MCP!
Got it — that clarifies how you’re defining the ecosystem. The architecture is certainly becoming much more connected than when we first spoke.
Thank you!
@UluPStudioManu — coming back to this because UluP has moved quite a bit since your original comment.
You mentioned that you were going to try FreshCtx against one of your LangGraph deployment tools. I never saw the result, so I don't want to assume you actually ran it.
But I noticed you've since shipped UluPy/MCP and are letting AI assistants create projects and manage tasks inside the UluP ecosystem.
That is actually more interesting to me than the original synthetic FreshCtx test.
Would you be interested in taking one real UluP MCP action — for example, an agent modifying a project or task — and putting FreshCtx at the pre-action boundary?
Then I'd like to run the same bounded action through Revera separately to see what we can independently establish actually executed.
I'm not looking for credentials, customer data, or access to the rest of UluP. One safe action path is enough.
FreshCtx:
https://github.com/Hyperwise-LLC/freshctx
Revera:
https://reveralab.com
If you're interested, I'd be happy to work through the first case with you rather than just send you another demo.
And if you did run that original LangGraph experiment, I'd still really like to know what happened — including if it didn't fit your architecture.