It's a data platform, sure, but it's not an OS, unless the definition of OS has suddenly shifted beneath me. That kind of thing "literally" happen to me.
Ontology is not some new "groundbreaking concept" invented by Palantir. I mostly hear it from philosophy and ML folks, but that's probably my personal bias.
Versioning data is not something that only Palantir figured out. I hear about this from different sources, like ML ops, but I've also dealt with it with creatives.
And I stopped there. I don't doubt that it all works together well, but this reads like someone's first experience with a single data platform, assuming that's the first one, true, data platform, and evangelizing.
Palantir themselves call it the Data Operating System, that's why I used this phrase as well, although it is of course not a real OS.
An Ontology is not a groundbreaking concept, you are totally right. "Ontologie" has been a philosophical discipline for hundreds of years already. It is also a big part of systems theory, which is party of most CS majors. The special thing is that Foundry got build with the Ontology as a core concept 10 years ago, when the whole industry was still using MapReduce. This gave them a huge headstart, which now allows users to build complex application on top of their platform.
Versioning data is not completely new, but the way it is integrated in the software is very impressing. If you look to the direct competition (Databricks, Snowflake, etc.) you're searching in vain for comparable features.
I am in fact evangelizing, because I think the data community can learn a lot from looking at this product. I myself use Databricks for the most part, and I love it. But I think the Foundry features I've listed are great as well and deserve some attention.
TL;DR: thank you for taking the time and reading the article! :)
https://en.wikipedia.org/wiki/Operating_system
Ontology is not some new "groundbreaking concept" invented by Palantir. I mostly hear it from philosophy and ML folks, but that's probably my personal bias.
https://en.wikipedia.org/wiki/Ontology_(computer_science)
Versioning data is not something that only Palantir figured out. I hear about this from different sources, like ML ops, but I've also dealt with it with creatives.
https://neptune.ai/blog/best-data-version-control-tools
And I stopped there. I don't doubt that it all works together well, but this reads like someone's first experience with a single data platform, assuming that's the first one, true, data platform, and evangelizing.