Palantir Platform

My main use case for Palantir Foundry  is pipelining and analyzing data there.

I replace the existing pipelines with Pipeline Builder in Palantir Foundry . I have various data flows and production of national reports, and I am replacing that using Palantir as part of an NHS Federated Data Platform. In terms of analytics, I use it to check data consistency and test it against what I have in other systems. People also use Quiver and Contour.

That is pretty much everything I have to add about my main use case or the way my team interacts with Palantir Foundry.

My background is in Databricks , and if I compare Palantir Foundry to Databricks , I see benefits of Palantir Foundry in that they make it simpler to configure clusters or at least to manage some infrastructure. If I think of Foundry  as being an implementation of Apache Spark  and compare that to Databricks, it is easier for an organization to use Foundry . I would also say that pipelining itself is more drag-and-drop style.

It is obviously easier to start with Palantir Foundry. I get more things managed by Palantir themselves. If I have a team with mostly SQL background and I want to move them to a Python, PySpark environment to use clusters, obviously using Palantir Foundry is an easier option than using Databricks.

There are pros and cons, obviously, regarding the features of Palantir Foundry. If I get stuck with the drag-and-drop nature of Pipeline Builder, it is going to be more difficult to migrate that to a different platform. From a Python coding perspective, even if I don't use much of that, I would say Databricks is probably better.

It is difficult to say how Palantir Foundry has impacted my organization positively. Palantir helped me migrate some data into the cloud. Whether they indeed impacted my organization positively is not clear because of Palantir's appalling reputation. So it is not that easy to say. If it was my choice, I wouldn't sign the contract with Palantir in the first place. I would probably stick to standard Databricks.

I wouldn't add more about the needed improvements, either on the technical side or regarding compatibility and integration.

Obviously the company's reputation needs to be improved regarding Palantir Foundry, or ideally, Palantir needs to get away from the appalling views on human rights and improve the reputation. Whether it can be improved, I don't know. This is not a technological problem; it is a problem of company image, so I wouldn't be surprised if the NHS actually triggers a break clause in the contract in February next year. That is not linked to the product itself. From a technical perspective, maybe to make Palantir Foundry more compatible with Databricks could be one option, or maybe more integrated with Azure . It is difficult to say because they might lose some of their competitive advantage in doing so.

I have been using Palantir Foundry for about nine months.

Palantir Foundry is okay in terms of stability. It gets sometimes occasional issues, but compared to Databricks, it is probably the same or may be better. However, as maybe one of early adopters, I get more technical support from Palantir, and I am maybe in a honeymoon phase. It is stable.

I don't know the answer to the question about Palantir Foundry's scalability, really, because I didn't test that. My assumption is that it is correct. I don't know the financial side of it, the cost. I can't judge that. To the best of my knowledge, it is not worse than Databricks.

Customer support for Palantir Foundry is okay. However, I am in this potentially better supported phase by Palantir Foundry. I am not admins on my tenant. What I have in the current environment is that Palantir Foundry themselves are running my tenant and configuring clusters and doing things that I need, simply because I don't have permissions. So when I get to that point of being admins on my own tenant, then I may be able to provide more information about that.

In the current organization, I used legacy on-premises and cloud databases, and I used and still use Microsoft Azure  before I switched to Palantir Foundry. I haven't fully switched yet, but the decision to overall switch was based on the momentum to go to this FDP, a Federated Data Platform. I believe it was a financial incentive to do so because from my organizational perspective, I am not paying for cloud space that I use in Palantir. I only pay for computing, but even that is probably covered by a bulk contract.

I didn't purchase Palantir Foundry through the AWS Marketplace  because in my case, it is a part of the contract between NHS England and Palantir. I don't know if they procured it via AWS Marketplace  or not.

My company does not have a business relationship with this vendor other than being a customer.

I haven't seen a return on investment with Palantir Foundry. I wouldn't even see the financial data. It is very difficult to judge for me. That is why if somebody would ask me whether Palantir Foundry in the NHS is value for money, it is difficult to answer that question.

I have no experience with pricing, setup cost, and licensing for Palantir Foundry because for me, essentially, it is free. NHS England pays for that as part of their procurement process. That is why I can't answer this question. But in terms of getting a contractor to work on that, I would probably say it is more expensive because there are fewer people with that skillset compared to, say, Databricks or Azure.

I didn't choose Palantir Foundry; I wouldn't choose them. If it was my choice, I would probably go for Databricks or even stay with Azure and try to see if I could use Spark there even without Databricks.

First of all, I would advise others looking into using Palantir Foundry to ask themselves if they want to use it given the reputation of the company. You don't need to use Palantir Foundry per se. The second consideration would be whether you want to use Databricks or any other implementation of Apache Spark . It would be interesting to see if you prefer the drag-and-drop nature of Pipeline Builder as opposed to, say, notebook structure of Databricks. So it might be a choice. I would probably say talk to your data engineers and ask for their opinion. Take that into consideration as well. Factors to consider include if you implement Palantir Foundry as what they consider a default option, you are likely to be very entangled into the product. It would be difficult to decouple in the future. That is why they are very sticky. That is probably one of the issues the NHS will get with the product in the future. My overall rating for Palantir Foundry is 6.

There are several use cases that we are working on with Palantir Foundry . The first thing is for data model creation for all our data engineering pipelines. That is one use case. Palantir Foundry  also has an ontology, more of a semantic layer, so that we can directly hand over the data model to the end users. That is another use case that we have, creating the semantic layer ontology. Recently, we have started working on some AI use cases as well. Palantir Foundry has very good wrappers such as AIP Agent Studio and AIP Logic, where you can choose any model and build your own chatbot or any AI function or generative AI function. These are a few use cases we are working on.

I work with different types of data in Palantir Foundry, including structured and unstructured data. We process PDFs and Word documents, but I have not worked on any use case with video and audio, although there are a few teams in our company that actually process video and audio as well. When it comes to textual information, I have worked on several use cases, and Palantir Foundry has made it very simple. There are some built-in functions, and you can also use Python libraries if you want. Additionally, there are no-code tools to parse unstructured information.

Based on my huge experience with Palantir Foundry, I find that starting from the data connection to the end user application, there is a tool for everyone. For example, I do a little bit of coding, and even for me, if I wanted to do some enhancements, I can do it and build specialized applications on it. If there is a person who is from a non-coding background, such as a business analyst, they will have some application to do that as well. For example, I have to build some pipelines on Spark, which needs to use several rows of huge data. There are also no-code tools available which are as good as Spark, not in terms of data management, but at least in terms of learning and ease of use. A business analyst can use that and prepare the same type of pipeline. For everyone, there is a tool or application in the platform.

We use collaborative functionality in Palantir Foundry. Initially, when I started, the collaboration function was not as matured as working on a general open-source application or something else outside the platform. Now it has matured a lot. Starting from the initial dataset, you can create branches up to the application level as well. This is a very good enhancement that Palantir Foundry has made in the last year or so.

The main benefits that Palantir Foundry provides for me as an end user are that everything is in one place. You can use multiple tools, and initially, there may be a little learning curve, but after you get started with it, you will find many advantages. I feel the advantages are on an organizational level. The main advantage is you can decentralize the analytics, and you will have everything in one place, so that you do not need to rely on multiple departments working on different tools. If your organization adapts to Palantir Foundry, then it is a totally different thing, because I see that advantage. I work in an internal audit team as a data engineer and data analyst, and sometimes we need data from different departments to do analysis. It became much easier for our organization because all the data is in Palantir Foundry. That is the main advantage that I see.

Regarding points for improvement for Palantir Foundry, I see that they are improving day by day. In the last one to two years, I have seen many improvements compared to the two years that I have worked on Palantir Foundry. There are many things that come up, but a few things are not intuitive enough. Now that we are in this AI phase, Palantir Foundry has created some wrappers around the models, allowing us to create using a no-code application, chatbots, and LLM functions. The problem is that interaction with outside applications can be difficult with the current setup that Palantir Foundry has. There are ways to do that, but it is not that intuitive, which is what I feel.

I have been working with Palantir Foundry for four years now.

For stability, I would rate it nine out of ten.

For scalability, I would rate it ten out of ten because you have a lot of flexibility. You do not need to worry about spinning up machines as you would with Azure . Here , you can select what memory or configuration you want. It is very easy, a ten out of ten.

I am not sure in general about the technical support, but at least for our company, it is very fluid. Whenever Palantir Foundry introduces a new product, the Palantir people come and train us on new applications, so I would rate that at least a nine or ten.

The setup process for Palantir Foundry is not complex at all. It is very simple. For me, the organization has already adopted Palantir Foundry, so I just need to deal with security policies that are already set up. I can directly start with sourcing the data and get started with the transformation. It is very easy for me, but at the organization level, I am not sure how difficult it is. Since we are in an insurance company, we have a lot of regulations, and I think there are many security policies in place, so all that setup is managed centrally for everyone in the organization.

Regarding pricing for Palantir Foundry, I am not entirely sure about the exact pricing because it is centrally managed by the organization.

When comparing Palantir Foundry with its main competitors on the market, I would say Azure  and AWS  are competitors because they offer several functionalities. However, Palantir primarily supports data management. I feel they are at the same level as those other cloud providers. I have also seen Microsoft Fabric , which has good pipeline building capabilities, but I am not sure about its AI capabilities since it is connected to Azure.

The visualization part in Palantir Foundry works for me at least if I want to see how the data is structured and for an initial analysis, but I would say it is not as matured as Power BI or Tableau in the market. Compared to Power BI and Tableau, it is not that mature. You can do a lot of things, but UI-wise, I feel Power BI and Tableau take precedence over what Palantir Foundry has. Palantir Foundry has it, but it is not as mature as those applications.

I am using some data integration features, including the in-built data integration feature in Palantir Foundry. We have several external sources apart from Palantir Foundry, so our application data is stored there. We use the data connection application and bring in all the data there.

I am not entirely sure about my level of satisfaction with the functionality of Palantir Foundry, but I think it is good. My overall review rating for Palantir Foundry is eight out of ten.