Palantir Platform

My main use case for Palantir Foundry involves building data pipelines, creating workshop apps, and constructing Gaia maps.

Another example of my main use case with Palantir Foundry is obtaining different data sources and combining them so that they can be visualized either in a workshop app or a Gaia map.

The unified picture is important for improved collaboration and decision-making in my organization, as that is the ultimate goal of a tier one organization in the Department of Defense and it is crucial to communicate to lower echelons effectively.

The best features Palantir Foundry offers include the ability to bring in multiple data sources into one spot and also host models that I can either bring or models Palantir already has access to, then combine them into a global ontology.

Combining data sources and hosting models in Palantir Foundry has helped my work because it is convenient to work in one environment rather than moving from one application to another, as Palantir Foundry allows for that one-stop shop where I can accomplish much of the work.

Palantir Foundry can be improved with better documentation, more robust training, and enhancements for working through transformations that are not accepted by the ontology. Additionally, the connection between Foundry and Gotham is not clear, and managing objects in Gotham lacks good documentation and training, leading to frustration. Using a regular database with a third-party application might provide a solution without being tied to the ontology.

Another drawback of the ontology is that it creates an additional step along the provenance of the data, which can slow things down or change what that data actually is once it reaches the end user.

Always having to work with a Palantir representative creates severe bottlenecks and increases costs, making it desirable for me as the end user to perform tasks without constant requests for support.

I would like to see a reduction in the need for field service representatives from Palantir, and I hope for a more intuitive architecture that makes it easier to find things and perform tasks without a high learning curve.

I have been working as a data scientist for six years.

I find that Palantir Foundry is stable sometimes.

The scalability of Palantir Foundry seems to be fairly good, considering how many users we have. It still operates well without significant lag in performance, so the scalability seems to be acceptable.

The customer support can be frustrating, depending on where I am working from, especially if the demand signal needs resolution from a Palantir representative.

We did not use a unified solution before.

My general impression is that it has not paid for itself yet, as it is a very expensive platform to use and the government is still fairly early in utilizing Palantir products. I would say that we have not received a good return on investment yet.

I did not evaluate any other options before choosing Palantir Foundry, as the choice was not mine to make. I was not responsible for selecting Palantir.

My advice to others looking into using Palantir Foundry is to seriously consider the cost of using it and whether you are comfortable relying on a Palantir representative to complete your work or if you think you can manage without any Palantir representation. Additionally, consider if your solution can follow a different path and make a comparison. My overall rating for this product is seven out of ten.

My main use case for Palantir Foundry is to modernize the data infrastructure. One of the modernization projects I have worked on involved getting all the telemetry data collected from IoT devices that had been sitting in the field and then streaming it to Foundry while using the AIP capabilities to perform predictive maintenance and forecast performance degradation of the metrics. This allows the AIP agents to send out remote fixes to address the actual issues.

Palantir Foundry helps with predictive maintenance and forecasting performance degradation by providing a layer of abstractions so that I do not have to worry about piecing together all the different frameworks. Rather, everything is integrated beneath Foundry and the AIP. I can focus on the data part, integration, and data integrity, which means I worry less about modeling and optimization.

In my recent project work, I have been extending all the AIP agents to derivatively send remote fixes. Rather than keeping autonomous operations confined within the platform, the agents can now interact with the real world to fix issues or conduct extended analysis so that the issue can be briefed in the ontology.

Palantir Foundry's best features include AIP, specifically its AIP capabilities. What stands out to me about the AIP capability is how well the data is tightly integrated, allowing me to ingest the data and then hydrate my ontology with context-rich data. Beneath this layer, the ontology creates its own semantic layer so that I do not have to connect all the dots. Rather, the AIP agent itself can look at the complete ontology and has its very own access, so I do not have to be feeding anything specific. Instead, I can give complete connected dots to my AI agents.

Palantir Foundry has positively impacted my organization by enabling us to gain traction from different industries and different companies across various sectors. Since PwC operates as a service-based company, we can pull out massive deals from those companies across various industries, making this a positive service implementation I have noticed in my company.

It has definitely increased the project delivery timeline, so now it does not take weeks or months to deliver a project but rather just days for the development efforts. This allows us to look ahead and spend more time with the business on actually understanding the problem rather than spending most of the time developing the solution itself.

Palantir Foundry could noticeably improve in providing visibility over the different layers beneath Apollo or the platform itself. Whenever an issue arises with a pipeline or an AIP agent that runs away with all the tokens, I do not feel enough visibility beneath the layers to dive deep into tracking the issue and then mitigating it.

The problem with the AI capability is that whenever I spin up an agent that goes and drags documentation, I feel less control over its actions. Since everything is tied together in the ontology, I really have a less structured and integrated way that I can intervene.

Customer support should definitely be a concern, especially for the dev tier account I have been using, while for a corporate account, it is pretty good.

I have been using Palantir Foundry for three years.

Palantir Foundry is generally stable, though sometimes when the data gets finicky, the Palantir pipelines or the ETL abstraction that the pipeline has breaks, making it hard to decode all the metrics and trace back the error.

I have not faced any issues with scalability, especially during long-running compute. However, sometimes it depends on the region where the subscription is deployed, which might lead to some temporary degradation. The issues usually get fixed within an hour or so.

Customer support should definitely be a concern, especially for the dev tier account I have been using, while for a corporate account, it is pretty good.

I did not previously use a different solution and was fully utilizing open-source frameworks and languages.

The setup cost and licensing are all simple, and with the documentation, I can literally navigate through a series of steps and then set up my own organizations.

Palantir Foundry has dramatically helped us in terms of project costing because earlier we had our own React developers team from offshore. Now with the AIP capabilities launched on the platform, we have completely avoided the need for a dedicated team. This has been very helpful in terms of cost management and reducing team size.

The pricing is a bit on the higher side.

Before choosing Palantir Foundry, I evaluated Azure Foundry. Since it was under development and in its early stage at that time, Palantir Foundry was beating it in its own game and was way ahead of Azure.

The accuracy and reliability of Palantir Foundry's AI output is pretty great. All those aspects are good, especially the documentation, which is so good that I can literally debug myself without looking for a long video that requires extended viewing time.

My advice to others looking into using Palantir Foundry is to get hands on with the platform and explore all its applications and the products that are available, as it is going to save a lot of time and money. I would rate this platform a 7 out of 10.

In my job, I use Palantir Foundry exclusively to create multiple dashboards. For example, I use Palantir Foundry to create a dashboard corresponding to the visualization of many charts by extracting the dataset, which is Skywise, putting this dataset in ontology, and using the different tools in Palantir Foundry. This is my typical use case in my job.

My last dashboard created with Palantir Foundry is regarding the Project Speed Project Dashboard, which helps analyze more programs because this dataset comes from Skywise, where my principal customer is Airbus. This project clarifies all the X-tracker, enabling tracking of multiple defects in programs such as the A320, and visualizing all action plans for non-quality across multiple programs. This is my first job for the dashboard speed, where I also plan to add, modify, and delete actions we want to track including all performance analysis for the high to left performance.

The best feature that Palantir Foundry offers in my experience is the Ontology Manager, which stands out to me because it allows us to see if we have the write-back dataset to understand what to add, delete, or modify in our dashboard and it displays our modifications in materialization, which is very good. Another aspect in ontology is that we have the possibility to update manually and see changes very quickly, which is a good feature that I apply and use in Palantir Foundry.

The Ontology Manager has helped me create an object or action, for example, using TypeScript, which is new for me, and it allows me to point to the Ontology Manager or the object type in the slate very quickly.

The Ontology Manager positively impacts my organization across all projects because it incorporates new technology and features that can be applied globally, making the impact on my work and organization very high.

I cannot provide specific outcomes or metrics on how Palantir Foundry has made a difference because in all my projects, I am the only developer and do not interact with other developers, only interacting with the customer, who is not a developer. Thus, I cannot see the difference at this time, as I am the sole developer on all my projects.

I want to pass the certification of Palantir Foundry because it is not easy to find the information regarding this certification, making it not accessible for many people, which I think is not good. If it is possible to plan for accessibility to this certification, it would greatly benefit many individuals.

I have nine years of experience in Palantir Foundry, which I used during my first internship and during my master's degree at the University in Nice Sophia Antipolis.

If I pass the certification, it would be the best thing for me as a Data Engineer, especially the Data Engineer Professional Palantir Foundry certification, which I consider important for my career.

I always take time to explore all aspects of Palantir Foundry, including the Ontology Manager, object set, object viewer, and object explorer, which I find valuable. Palantir Foundry has improved with the generation of AI, and I think the governance and security are both good things to have in Palantir Foundry.

I find that the AI capabilities of Palantir Foundry provide great accuracy and reliability, comparable to tools such as ChatGPT and Google Gemini, indicating strong output in terms of accuracy and reliability.

My advice for others looking into Palantir Foundry is to pass the certification, which I believe is very important to demonstrate that one's experience is applicable and valuable in exploring everything that Palantir Foundry has to offer. I rate Palantir Foundry a nine out of ten because I have a good experience and find working in Palantir Foundry very easy, as it offers many possibilities for growth and cooperation with people from all over the world.

I work in an organization which uses Palantir Foundry, not as a customer, integrator, reseller, or consultant.

Our main problem statement with Palantir Foundry was that we have our data in corporate systems, primarily SAP-based, and many other systems managing HR, logistics, finances, and program activities, all working in silos without communication. Whenever we need to integrate something, it has to be done outside those systems, which could involve a lengthy, customized solution process and maintenance hassles. Updating the data becomes problematic as it turns static once extracted, requiring repetition for new updates, which compounds errors and effort across 120 countries and multiple locations. There was no central truth to the data, leading to discrepancies as everyone applies their own algorithms. This ongoing issue is rampant across organizations; I recognized it since joining in 2001, noting that other solutions like Tableau or Power BI merely mask the problem rather than solve it. Palantir Foundry was clearly the solution we needed; it has upgraded our thinking, decision-making processes, data handling, and overall data literacy. Now, the democratization of data occurs as everyone learns to access and analyze it uniformly. This shift has fostered new habits in data sharing and reporting while making previously theoretical efficiencies actionable.

What I appreciate most about Palantir Foundry is its focus away from mere reporting beautification; organizations historically prioritize chart aesthetics and transitions. Tableau exemplifies this with flashy features that, while entertaining, lack substance. The real value lies in key numbers and easy complex calculations, traditionally requiring extensive configuration. With Palantir Foundry we have eliminated the complications of such custom solutions, and it synchronizes data across systems effortlessly. Whether I am handling 10 rows or 10 billion, it processes the data efficiently, employing Spark for computations based on data volume. Moreover, the integration of foundational technologies into Palantir Foundry allows users to work with Python or SQL, maintaining a universally recognizable skillset. Even if you are new to these languages, learning them within Palantir Foundry equips you for future career shifts. While Palantir Foundry is becoming somewhat restrictive, it fundamentally educates users on widely applicable skills.

I believe one significant enhancement for Palantir Foundry could be making it more accessible for managers and decision-makers who often lack time for in-depth analytics. Currently, analysts handle the in-depth analytics and present information to managers. Bridging this gap means making it effortless for users to quickly open the platform and find the information they seek, which could improve with the introduction of AI assistants. Instead of needing to navigate complex queries, users can simply ask questions and receive the necessary data, making managers happier with Palantir Foundry's capabilities.

I have been working with Palantir Foundry for about seven to eight years.

My impression of the stability and reliability of Palantir Foundry is positive; it is well managed and robust, and I have not observed breakdowns. Palantir appears to invest significant resources in support at various levels, allowing problems to be escalated effectively, despite occasional bureaucratic delays. The platform itself is well-crafted, and improvements are continually rolled out.

The scalability of Palantir Foundry is immense, suitable for any industry and can expand to encompass various departments and fields that process data or workflows. It has applications across banking, insurance, logistics, supply chain, and engineering. However, I observe a potential drawback concerning market trends; as software becomes more accessible with individuals developing applications independently, there is a risk posed to Palantir Foundry, which combines multiple core functionalities. Although its significance remains vital, emerging competition can impact its customer base. Microsoft and Anthropic are both exploring ways to capture part of Palantir Foundry's market share by offering solutions that simplify tasks that would otherwise require substantial engineering resources.

I perceive the technical support for Palantir Foundry as standard and adequate; I interact primarily with the deployed engineers, though I cannot distinguish which ones are from Palantir or from us.

I have not evaluated other options before choosing Palantir Foundry because it had no competition seven to eight years ago. Other solutions such as building your own pipeline through IT were difficult to support and upgrade, often failing to connect to APIs. Once organizations grow larger, specialized units emerge, isolating business users from technical expertise, but Palantir Foundry integrates both, empowering technical data analytics for business users. It is increasingly relevant today with the rise of AI, as those sticking with Excel will struggle compared to specialists who can drive efficiency through data. Palantir Foundry enables a mutual understanding between technical and business perspectives. Competitors such as Tableau, while popular, provided only dashboards without addressing core issues. The same goes for BO and Power BI. Recently, Microsoft Fabric might be a real competitor, but I have no experience with it.

I participated in the initial setup of Palantir Foundry from a business perspective, particularly in setting up the ontology. Connecting to corporate systems was handled by Palantir engineers along with our specialized technical staff, ensuring governance regarding security, safety, and the ingestion process. We created the first ontology and applications to demonstrate Palantir Foundry's capabilities. Additionally, we developed APIs that interfaced with legacy applications for data crunching in Palantir Foundry. We implemented code repositories to ensure robust production-level pipelines. As early adopters, our involvement predates Palantir's public launch, making us a key part of its development journey.

I think there are quantifiable efficiency gains from Palantir Foundry, but I cannot specify exact numbers; we have verified that it helps in saving costs.

Navigating Palantir Foundry is straightforward for someone with a technical background, and the provided documentation and learning platform, learn.palantir.com, support various use cases with hands-on experience of new features. The AI assistant, although not perfect, offers assistance for troubleshooting. One of Palantir Foundry's strengths lies in its adherence to industry standards, allowing familiar practices from other applications to be applicable. Its version control, Git usage, and standardized charts provide multiple support avenues for users. I would rate this review a 9 out of 10.

I used Palantir Foundry in a different role prior to my current role; however, I do not use it in my current position.

In my previous role, my main use case for Palantir Foundry was data extraction, ingestion, and cleansing for use with machine learning models. A specific use case involved extracting data from multiple public health sources of patients and providing healthcare providers with legal data and logistical data about healthcare providers to analyze which healthcare providers were guilty of fraudulent billing.

Other use cases included ingesting veteran data for different kinds of machine learning using clinical data for veterans; however, the primary use case most recently was fraud detection for prevention of fraud, waste, and abuse in federal health agencies using data from NPI, National Provider Identification, for various providers along with state licensure data for fraud detection.

Based on my experience, Palantir Foundry is extremely easy to learn and adjust to since it is primarily a closed system, meaning that all the functions for data ingestion, data cleaning, machine learning, and related tasks can be performed from within the same system. This closed nature made it easier for my junior engineers to use the system for working on the actual work of ingesting the data.

The Ontology was one of the more helpful features alongside ease of use. Palantir Foundry has impacted my organization positively; the general output is comparable to other systems, but the value came more from the speed of querying and building tools to extract and transform the data.

The main benefit was having something that was faster to get to the result set, resulting in less manual work and fairly automated processes. The Ontology system was helpful for our teams.

This system needs more powerful tools for the power user; I feel the system is very well designed for the introductory level but could have finer-grained controls for data engineering experts and machine learning experts at the power user level.

I have been working in my current field for eight months.

There were no issues with stability in our experience; we had no problems with that.

We use Palantir Foundry for fairly large data sets; however, we did not scale to full size since we were building POCs and supporting the federal government in their use cases, and we definitely did not have any issues with scalability on our side.

I cannot comment on customer support since I did not have to reach out to customer support for anything during my use of this system; however, I can comment on training and staff technology partnerships, as the Palantir trainers at the federal government were extremely friendly and very helpful.

We did not have a different solution for this particular use, as the federal client was using a different use case and using Palantir Foundry from the outset; ours was a different use case for which Palantir Foundry was the first technology selected.

Overall, I supervised a team that used Palantir Foundry, and the primary benefit here was that there was faster adoption of the technology and faster time to results, primarily with the technology enabling us to have quicker POCs, proof of concepts, using the data system.

I would say that there was significant time savings with Palantir Foundry; these may have translated into money savings, but typically we were more focused on the time savings.

I was not typically responsible for pricing, setup cost, or licensing, as I was primarily the leader of the project that implemented the data ingestion, cleaning, and machine learning tasks and led the team; pricing and infrastructure setup would have been handled by other technical leaders in other domains.

We did evaluate other options, but Palantir Foundry was the natural fit for this particular use case, so we did not evaluate other options such as Databricks particularly seriously.

My advice to others looking into using Palantir Foundry is to ensure that new users are taught fundamentals of data engineering, data ingestion, and data science before they go into the tool, so they do not have to learn it inside the tool but already come in knowing what to expect; learning from within the tool would create a more restricted learning system, and while the tool is powerful, it should not be used for learning systems but for developing systems. I provided this review a rating of eight.

My main use case for Palantir Foundry is working for a client, Airbus, where I work on the datasets. Currently, we are working on multiple services such as Code Repo, Slate, and Workshop.

We have Contour, where we analyze the data and find the anomalies. We look into the graphs, plot graphs, and investigate how the data is behaving. If the data is not up to date, we investigate that on Contour. In the data lineage part, we backtrack to the point of failure where the failure point is occurring.

I have worked on a project called CMA, wherein we have a Slate application. The dataset is a writeback dataset wherein we get the data as a writeback phonograph sync, and we get the output from the users, process the data, transform the data, and export the data to FTS+.

The best features Palantir Foundry offers include clickable outputs that we can easily get, such as the summary of a column straight away. By the click of a button, we can get the column names, and even if there are N number of columns, we can have the column count, row count, and multiple other data specifications in the tool itself. That is a very good thing I have seen, and on the Slate part, we have seen functions, queries, and objects that are very good.

These features help me in my daily work and improve my workflow by automating things and automating our day-to-day job. We use queries, functions, and all. Code Repo also works in this case, and data lineage along with multiple other capabilities are very useful to us.

Palantir Foundry has positively impacted my organization, especially Airbus, as we are dependent on Palantir Foundry for processing the data. We are very much more dependent on the tool.

It has especially improved data accuracy for us, wherein we use humongous data to build applications based on that particular big data, which is quite good since we have a tool Palantir Foundry.

Palantir Foundry can be improved by providing training to the people who are working. I feel there is a lot of training that needs to be provided to the developers of Palantir.

The way you handle the products is where improvements are needed, especially the training part. To add features, we need to read the whole documentation, which is time-consuming and wasteful. Providing more training and more videos on YouTube would help.

Regarding Palantir Foundry's AI capabilities, I think its governance and security could be improved. Palantir Assist could have more interactive ways, and I feel the interface where Palantir Assist is provided is not that good.

I have been using Palantir Foundry since one and a half years ago.

Palantir Foundry's stability is sometimes good and sometimes not; there are blunders and issues.

The scalability of Palantir Foundry is awesome. We can easily process and build applications on humongous data.

Customer support is nice. I would rate customer support an eight, maybe.

I have not used any other solutions previously.

I have seen a return on investment. Time is saved once we try to build applications on big data where the capacity of handling the data is awesome in Palantir, so I appreciate that.

The pricing is quite high, and there are obstacles in licensing that could be made more flexible.

I did evaluate Azure Databricks before choosing Palantir Foundry.

The accuracy and reliability of output from Palantir Foundry depend on the model. The advice I would give others looking into using Palantir Foundry centers around access and pricing, as these are the two things that need improvement. I hope we go with AWS for our cloud provider, as we have not worked with other clouds. I would rate this review a seven overall.