> It is essentially an interface that sits on top of existing data sets and displays data to users for analysis, helping to identify connections otherwise impossible to find. Users do not have to use SQL queries or employ engineers to write strings in order to search petabytes of data. Instead, natural language is used to query data and results are returned in real-time. It is not designed to do any single thing, its main strength is that it is flexible and powerful enough to accommodate the requirements of any organization that needs to process large amounts of both personal and abstract data. This makes it more useful for managing HUMINT, or intelligence from human sources, than SIGINT, or intelligence from signals.
The problem with Palantir is that everything it is said to be doing is so vague that it could just be colorful dashboards based on not particularly great or new data. For example, this is how the Bloomberg story you linked to describes it:
> Founded in 2004, Palantir is used by dozens of federal, state and local law enforcement agencies to aggregate far-flung data, find patterns and present results in colorful, easy-to-interpret graphics. Its use by police in Los Angeles, Chicago, New Orleans and elsewhere has raised ethical concerns about the potential for unfairly targeting minorities.
I'm not arguing whether Palantir is living up to its contracts. Just that it doesn't seem to produce or have the unique and expansive kind of data that Facebook, Google, and Amazon have. Though I guess there's nothing that prevented them from doing a massive data pull from FB's API, in the same way CA managed to do it.
FWIW, not everything about Palantir is necessarily a secret. You can see some descriptions of what they offer and at what price via various FOIA requests that have been done: https://www.muckrock.com/foi/list/?q=palantir&status=done&us...