Palantir Foundry Reviews

My main use case for Palantir Foundry is from the data engineering perspective.

A specific example of how I use Palantir Foundry for data engineering involves raw data stored in Redshift AWS, which we are using those tables in the form of a dataset in Foundry. We are ingesting that data into Foundry and using it for cleaning purposes. After cleaning the data, we create Ontology objects and use them for operational applications in the Workshop.

One of the use cases that I found with Palantir Foundry is when I worked on the supplier scorecard, which is dedicated to understanding supplier reviews based on the goods supplied. The company assigns ratings to their products through a supplier scorecard, providing scores to their suppliers. We used multiple datasets and created objects, adding our own logic in the Code Repository to check supplier goods by percentage and count, generating aggregated values in the Workshop app. Based on these parameters, business management can make decisions and take actions to update the supplier's score.

Palantir Foundry offers great features, including data connection specifically for ingesting data from various third-party servers like AWS or Azure, with around 250 plus data connections available. Additionally, it includes Pipeline Builder, one of the best ETL tools for transforming data from raw to gold layer data, following a medallion architecture of bronze, silver, and gold. In more complex use cases, the Code Repository offers a fully code-based solution while Pipeline Builder serves as a no-code, low-code tool; my preference leans towards Pipeline Builder for data refinement.

For data analytics, it features Contour, allowing for data analysis, and ontology objects for creating links between multiple objects with actions for CRUD operations throughout the Workshop. It also has Quiver for exploring objects using AI tools, enabling business users to ask questions in their native language, which Quiver converts into queries for report generation. Another significant feature is AIP Logic, akin to agentic AI, processing existing data with multiple AI models. AIFTA is another cool feature that needs only a prompt to handle various tasks, such as creating ETL pipelines based on raw data stored, selecting to create the pipeline in either Pipeline Builder or Code Repository as needed, and also supporting object creation at the branch level.

In the Pipeline Builder, we can use Databricks as a compute profile, which is one of the coolest features. The OSDK is a new feature that allows creating custom UI pages using React or Angular, fetching data through API, should the business be unsatisfied with existing widgets in the Workshop. There is also MCP Hub, which uses Model Context Protocol to operate Palantir Foundry from a local machine using LLMs, generating and deploying code efficiently.

Palantir Foundry has positively impacted my organization through multiple use cases, such as warranty data refinement, where previously we struggled with identifying the number of claims related to specific products. Analyzing the data helped us craft a Workshop application that tracks claims by country and product, enabling report generation for management action on defective products.

I believe Palantir Foundry could improve by introducing a tool to restrict object-level creation to specific people, such as developers. A dedicated application could streamline requests for access to data across different organizational verticals, enabling better tracking of costs associated with specific use cases and improving identification of data access requests.

Regarding documentation, I find that when I face issues, the outdated documentation is not helpful; for example, while trying to create a webhook to fetch SharePoint metadata, I found available resources lacking relevance, needing significant updates to assist users properly.

I have been using Palantir Foundry for the last four years.

Palantir Foundry is stable.

Regarding scalability, if you have billions and trillions of records, Palantir Foundry accommodates ETL pipelines with a dedicated compute profile that lets you choose from several sizes, whether small, large, extra-large, or custom to fit your needs.

Customer support is really good; when we encounter issues, raising a ticket with a screenshot leads to responses typically within a week or twice a month, depending on their organization.

Previously, for data engineering, we used Databricks; however, it lacked the capabilities we found in Palantir Foundry, which allow for analysis, reporting, and automation without needing to implement additional functions such as AWS Lambda. I appreciate that Palantir Foundry offers dedicated automation tools significantly simplifying processes.

We have seen a return on investment, primarily saving money on developers. With traditional development requiring many specialized roles, Palantir Foundry allows us to operate efficiently with fewer personnel, often enabling just a couple of front-end developers to manage our processes, thus noticeably reducing time and costs when completed effectively.

My experience with pricing, setup cost, and licensing has not been too overwhelming; I worked closely with a management colleague who explained how they check for cost based on user activity and individual vertical usage.

I primarily evaluated Databricks, which I found lacking compared to Palantir Foundry's robust offerings.

My advice for others considering Palantir Foundry is that it delivers an ecosystem eliminating the need for third-party applications, greatly simplifying tasks without requiring extensive efforts in model training or other processes, making it a strong option for organizations. Security and data governance are also significant advantages. I have covered everything I know and have used regarding Palantir Foundry. I would rate this product a ten 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.

I have used Palantir Foundry in multiple cases, such as creating data pipelines and ingesting data into Palantir Foundry from various data sources, including structured, unstructured, and semi-structured data. After ingesting into Palantir Foundry, I have cleaned the data using PySpark and code repositories. I have written Python and PySpark scripts that clean the data using various transformations, schema changes such as converting boolean and string fields to boolean, data type changes, dropping unwanted data, and filtering the data. I am building end-to-end pipelines in which I have joined and integrated multiple data sets. I have also applied various health checks, scheduled jobs using Cron expressions, and created ontologies, including actions such as create, edit, and delete. On top of ontologies, I have created Workshop applications for the UI perspective and multiple kinds of UI applications. I have also worked on the Slate part. Regarding industries, I have worked in the aerospace industry, healthcare, and the gas and energy sector, with major clients in three industries.

I appreciate multiple aspects of Palantir Foundry. I start with the clean architecture and clean UI. Regarding the coding part, I use code repositories where I can create multiple Python and PySpark scripts and test them. Palantir Foundry has data lineage that shows my data pipeline in a graph, illustrating the data flow from data nodes. If I have a thousand datasets and a column called test, for instance, and I want to check which dataset this column is coming from, I can check it easily in the data lineage by typing the column name. I can debug my bugs very easily using it. Palantir Foundry has features to track daily pipelines, identify which dataset is failing, and track time since the last check, along with health-related aspects that I can monitor using that lineage.

On the Ontology UI part, using Workshop applications, I can create very interactive applications quickly and easily. These are very cool features in Palantir Foundry, and now with AIP, it is also very useful.

Palantir Foundry has a service called Resource Management, where I can track how much the architecture is costing and how much data I have in Ontology and datasets. This tracking is beneficial since the UI is very clean, and I really appreciate it.

Palantir Foundry could improve in several areas. I do not prefer Slate. I think Palantir has stopped developing Slate and is focusing more on Workshop applications, which they are developing rapidly. Slate does not have a branching concept where I can deploy changes. For example, if I have production and if I have QA or development, there is no feature to deploy changes from dev to QA or QA to prod. I need to do manual replication of my work in QA and prod. I think Palantir will drop that application in the future since they are not interested in Slate. Another improvement is needed in integration with multiple services. While Palantir is increasing integrations with other platforms, I believe more platforms should be added. Additionally, sometimes it is very hard to use Palantir APIs, and the documentation has very little information.

Slate is a UI perspective application in which I can write JavaScript functions. Palantir Foundry has a code sandbox where I can write HTML, CSS, and other elements. Slate is primarily used to create UI applications, but it lacks a branching concept for change deployment. Without this feature, I need to replicate my work manually in different environments. I believe that Palantir is not interested in Slate's development, and they will likely drop that application in the future.

I have been using Palantir Foundry for around five plus years.

I have not had much interaction with customer support. I have connected with the Palantir support team one or two times, but not often. Whenever I face an issue, I use Palantir AIP, which is very useful.

The support quality and speed depend on the contract. In one of my projects, I had a weekly call with the Palantir support team, which was very useful for addressing any questions or doubts regarding Palantir Foundry. However, in another project where there was no paid contract, the support was not good, and they were slow and not very helpful.

I have worked on Databricks, and similar to Palantir Foundry, I have written PySpark scripts in Databricks and created some UI applications, although that was a long time ago. I am not familiar with the current UI services in Databricks. I have worked on AWS Lambda and Glue, but not much. I have worked mostly on Palantir Foundry.

Whenever a client wants to initially come on Palantir Foundry, the process will be very easy. I need to consider multiple points in my mind when starting any project on Palantir Foundry. For instance, I worked with a major healthcare client who moved from Databricks to Palantir Foundry, and I helped them with the transition.

Regarding pricing, I heard about it, and until last year, the license was one million dollars. They have now increased this to four million dollars, which is high. From a pricing perspective, I have also worked on optimizing costs. I had existing pipelines that used multiple resources, which increased costs. I focused on optimizing our pipelines and code to use fewer resources, which means lower prices.

I have more than five years of experience overall, and from the start of my career, I have been working on Palantir Foundry. I am also a Palantir certified data engineer.

I use Palantir Foundry to ingest data and create visualizations for decisions.

My favorite thing about Palantir Foundry is the ability to utilize multiple different types of coding languages and the current updated documentation and help features that are in there have improved pretty significantly and make it rather friendly for use.

The scalability of Palantir Foundry is the part that I love the most. That is why I said that is where my primary focus is, trying to get data and then turn it into information and knowledge and understanding, and more importantly, getting decisions out of it. The scalability is phenomenal.

I think the things that I do not like about Palantir Foundry is not a Palantir issue so much as it is from my company side and what they have commissioned for and have not commissioned for.

With Palantir Foundry, it is a part of the user tools that they provide. Their AI Assist that they use is something I have found that sometimes I get better results for when I do need help and aid. I get better results going to outsourced AI assistance sites such as Gemini, because it seems like AI Assist, which my understanding is it is supposed to be searching and utilizing the documentation for Palantir Foundry, but sometimes it gets kind of confused in the capabilities and it will tell you, 'Oh, you can use this and do this.' And then when you try it, the system says, 'We do not support this.' Whereas I can go to Gemini and it will give me a workaround that will actually work.

I have been using Palantir Foundry for about four years now.

I find the stability of Palantir Foundry to be extremely stable. I have had only one time period where there was any downtime that I noticed, and it was for a very short period of time. When I say short period of time, I am talking within hours, not days.

I have contacted the technical support or customer support of Palantir a couple times.

Especially for what they are supporting and doing, I find the quality and the speed of the support to be extremely fast. I usually get within a day turnaround and the support staff are extremely knowledgeable and good at what they are doing.

Prior to using Palantir Foundry, I used various different SAPs and ERPs where I would actually have to export worksheets and then build my own. The closest thing I would be building would be databases that just resided on my computer, and I was not using anything like Palantir Foundry.

The initial deployment of Palantir Foundry was super simple.

It did not take me any time to fully set up Palantir Foundry because it was made available to us and it was basically already set up and rolled out by the time I got permissions to use it. All I had to do is basically create an account. Actually, I did not even do that. Someone else in my organization created the account for me with the initial setup of everything. By the time I logged in and went to use things, most of it was already initialized for me.

Palantir Foundry requires no maintenance on my end as it is taken care of by Palantir. The only thing I ever get, and Palantir already does it, is when there are certain upgrades to transformations that I have made, they will put in the upgrades, but it requires me to actually approve them and merge them in. It is minimal work on my part. Usually I can click it and just approve it right away.

I would rate this product a 9 out of 10.

Palantir Foundry serves as our data platform for the company, which means we have numerous use cases and business cases that cross all the different business groups, subsidiaries of the company, and also different support functions and business functions of the corporate. We have more than 200 use cases in the corporate because Forvia is a very big company. The main use case is to enable the data value and data product for our corporate and for our business.

The main purpose of the data platform is to have a good return on investment based on IT digital dependencies. From a business point of view, I will give you a good example of purchasing. For the purchasing side, purchasing has two types: direct purchasing and indirect purchasing. Especially for the direct purchasing part, previously, we could not know that all the purchasing data management was quite siloed. With Palantir Foundry, we break the data silo to make all the different data which comes from the purchasing department globally, which have acceleration with the data sourcing assistant and AI sourcing assistant, to help our business accelerate their purchasing business transformation and to achieve excellence in terms of purchasing goods price. This helps us, at the same time, to speed up for the purpose of time saving, and it helps our business to accelerate all the price transformation strategy with our suppliers. That is a good benefit.

Not only for the purchasing part, it is also for the processing side in the operation and for the industrial operation, because Forvia is a manufacturing company. We have many data use cases in the plant. Globally, we have 500 plants and factories globally, which have many critical operations on the factory plant side. For example, the predictive maintenance with the data coming from the shop floor from the plant side helps us to have a good level of understanding of the different machine statuses of the plant.

This is the data-driven enterprise strategy. Since five years ago, we started our data program and launched the data-driven enterprise. This strategy has changed our HR organization, meaning we need to apply change management to accelerate because we are facing the change of data and AI. With Palantir Foundry, it helps us to accelerate this change management in our corporate, which is quite positive.

Time saving, budget saving, and cost reduction are benefits we have experienced, along with accelerating decision-making for the target, because Palantir Foundry with the data is quite useful. It helps management make the right decisions in the market, especially in the current situation where all the competition in the automotive market is quite complex. With Palantir Foundry, it helps us have better benefits and better return on investment, and also accelerates the right decision in the market.

There are three good features which we have applied until today in Palantir Foundry. The first one is, of course, all the data product features from Palantir Foundry, with all the different data pipelines, which helps us to have end-to-end data product experience with Palantir Foundry. The second one, relative to the previous benefits about data product, is that we have a good level of data ontology, which is a data catalog that helps the business people to understand better their data in a functional way. The third part is the AIP usage, because Palantir Foundry has the AIP feature, AI platform feature. With AIP features, we could accelerate our AI transformation and also develop our own AI agent with Palantir Foundry.

Palantir Foundry needs two points for improvement regarding the data product. First, Palantir Foundry needs to improve their clear resume about their product features roadmap. Second, Palantir Foundry needs to have a closer connection with the enterprise corporate application, which means the business application, because big companies have a very huge ecosystem of business applications. In my personal perspective, I think Palantir Foundry still has some space to improve in integrating with the IT landscape of the corporate.

I want to say that Palantir Foundry is quite expensive. It is not so easy for budget review and budget transformation of the company, which is quite expensive.

In terms of my experience with Palantir Foundry, I have been using the Foundry product from Palantir for more than five years already.

It is stable.

The scalability is good, but we need to pay for the compute and the resource.

The customer support is fine. We have the Forward Deployment Engineer, FDE, with us on site, but once again, it is quite expensive for the daily price of the FDE engineer. I think we need to rely on classical support by using a ticketing system of Palantir.

We previously had Cloudera in the company as the data lake solution.

At this stage, it is fine.

We don't have a migration plan.

I cannot give you the details of the money saved because it is quite confidential. What I can tell you is that return on investment is quite good, but Palantir Foundry is quite expensive and it is difficult to have a good budget for Palantir Foundry. That is the reality.

At this stage, it is fine.

At this stage, it is fine.

I have two suggestions for other companies looking to use Palantir Foundry. First, you need to understand how Palantir Foundry integrates with your IT system landscape before choosing to use Palantir Foundry. Second, you need to define and design good governance for Palantir Foundry usage for your data platform. I have rated this review with a score of 8.

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.

I have been using Palantir Foundry for about two to two and a half years. My main use case for Palantir Foundry is data analytics.

I get asked to do a particular project for data analytics. I research Palantir Foundry for the datasets that I am looking for. Sometimes I create datasets from other datasets, and then I either export the file or create a report on Palantir Foundry.

I think you have to exercise using Palantir Foundry to better understand how it works. However, there are tutorials and AI assistance with Palantir Foundry, which makes things easier.

The best features Palantir Foundry offers include the Workshop, which is excellent. It offers a widgetized method of developing dashboards and reports. The AI assistance is very good for trying to shape your report.

The Workshop feature is an app that allows me to build a dashboard or a report on Palantir Foundry.

Palantir Foundry impacts my organization positively by creating a central repository of data sources from all the other sources of data, making it a one-stop shop for any sort of data research.

It has made things easier in finding data. Though as a caution, had I tried to search for this data the normal way through the normal source, I probably would not have been allowed access to said data. However, because all the data is being pulled into Palantir Foundry, it makes it easier for me to access data that I have been restricted from.

In my use of Palantir Foundry, many people can go in there and create datasets, save datasets, and share datasets. However, if many people make datasets of low quality or if they are using the same name for datasets, it can get very confusing. So, it does not seem like there are any sort of business rules when it comes to naming your dataset or keeping your dataset active, making it quite messy depending on who is accessing it and what they are doing with it.

I am pretty sure Palantir will get around to coming up with better business rules and cleaning up bad datasets. It is only a matter of time.

I have been working in my current field for about four years.

As far as I know, Palantir Foundry is stable.

I believe there is a team working on gathering more data sources for Palantir Foundry. I can request additional data connectors, but it seems so far there has not been much restriction or rejection for additional data connectors.

Customer support for Palantir Foundry has been pretty good so far; I have only had one incident, and I got a response back rather quickly.

Previously, I would have to pull Excel spreadsheets from various sources and then make a report out of Power BI. I still use Power BI, just instead of going to multiple sources, I go to Palantir Foundry as a one-stop shop for my data sourcing.

Palantir Foundry makes data reporting easier and reduces time in data gathering and reporting. What would usually take about a day and a half of data gathering and reporting, I have ultimately reduced to about 20 minutes.

I do not have any experience with pricing, setup cost, or licensing since it all comes with the company that I work with.

There were no options before choosing Palantir Foundry; it is pretty much promoted by the company.

My advice to others looking into using Palantir Foundry is to go through the tutorials that are offered. Take the time to go through the lessons, and if you have any experience in SQL, probably improve your SQL knowledge. I would rate this product a seven out of ten.

My main use case for Palantir Foundry is primarily focused on time series related information, visualization, and a few applications where the background involves AI.

I receive different sensor data through time series and apply business logic on top of that. With the aggregated data, I perform visualization according to business requirements. The logics and everything are implemented both as core native logic within Palantir Foundry itself.

This is the most common use case I work through. Apart from that, there are a couple of additional projects involving workflows where resource management needs to be handled. This includes resourcing schedules as well as job allocation.

Palantir Foundry is deployed in my organization as a public cloud only.

In my opinion, the best features Palantir Foundry offers are that it is not rigid and provides low-code, no-code capabilities.

The low-code, no-code facility allows people with less technical knowledge who have domain knowledge in a particular field to directly use the application and readymade widgets to prepare their applications in a much faster way.

Another advantage is the Ontology layer, which serves as a business layer. Once the data is set on the Ontology layer, it can be accessed across multiple divisions.

These are the two main points. The overall architecture is definitely very robust and handles both the velocity and volume of data so that end users do not need to manage these concerns.

The sensor data and applications built on top of Palantir Foundry represent the main advantage my organization is currently taking.

The widgets are pretty limited. While they continue to improve, the widgets remain limited. If you want to create customized applications, it will be difficult. Using their standard widgets and features works very well, but any kind of additional customization needed will be challenging.

There are many widgets for specific needs or specific ways to build applications, but not all of those widgets or features are available in Palantir Foundry.

I have been using Palantir Foundry since 2023 across different projects within my current organization.

Palantir Foundry is stable based on my experience.

I am not certain about Palantir Foundry's scalability because I am particularly on the data engineering services side.

The customer support for Palantir Foundry was good, but it can be improved.

I came across multiple data engineering solutions before using Palantir Foundry.

I think it is both a time-saver and enables better decision-making because the sooner we get predictions or anomaly detection, the more helpful it is.

Compared to other SaaS tools, Palantir Foundry is definitely a time-saver, though I do not have specific metrics to share.

As a whole product, Palantir Foundry is well secured. Even though there is a capability of integrating with external applications, if all your data resides in Palantir Foundry, it is quite secured and includes most governance and security measures.

Personally, I am not involved much with AI capability related work.

I am not certain which cloud provider is used for Palantir Foundry. As a developer, I am not much aware of which cloud provider is used for Palantir Foundry.

My overall rating for this review is 7.

My main use case for Palantir Foundry is for decision intelligence. It functions as an operational operating system that connects fragmented data into a single ontology for our insurance provider, utilizing principal and component analysis.

An example of how I use Palantir Foundry for decision inclusions or connecting fragmented data is through the core mechanism of data connections to building Pipeline Builder, then using ontology chain. Data connection ingests data from different disparate sources such as Zabbix, health signals from Zabbix, and PostgreSQL external APIs. Now living in separate systems, Foundry pulls this into a unified raw dataset layer without needing custom ETL glue code. I transform those raw inputs into clean, joined datasets, and the ontology is where fragmentation is truly resolved. Instead of querying disparate tables, every entity, device, and incident becomes a pipeline run that functions as an object type with properties and links to related objects. A network object, for example, links to its Zabbix alert objects, its health metrics time series, and its owner, all the exact type of multi-source join that I can rebuild, rather than writing SQL joins every time. The ontology makes these relationships first-class.

In our infrastructure data spread across Zabbix, PostgreSQL, and Zyco, the classic problem is that each system has its own ID scheme, update cadence, and schema. Palantir Foundry's ontology solves this by creating a canonical object model on top. I do not migrate or replace the source systems; I overlay a unified semantic layer. Every downstream consumer reads from that layer and not from the raw source directly.

The best features Palantir Foundry offers include the ontology, which is not just a tool. Palantir Foundry can collect data from an existing system without modifying those systems, and once the ontology is ready, I can create a business application, analyze data, or build AI models on top of it. Team report projects' speed becomes faster than using multiple separate values. The ontology is not just a data model; it is a semantic layer that represents real-world entities, relationships, and decisions that act on them. Palantir Foundry supports ingestion, transformation, semantic modeling, analytics, and operational application development in one platform. This reduces the need to switch between separate tools for pipelines, governance, and downstream consumption. It is well-suited to use cases where analytics output must be embedded into operational processes rather than limited to dashboards.

Using this semantic layer in Palantir Foundry connects the fragmented data that meant writing JSON logic repeatedly across dbt models, Snowflake views, and Airflow DAGs. Every new consumer, such as a dashboard, a report, or an API, needed its own interpretation of what a device, incident, or transaction meant. Schema changes broke downstream consumers silently. Governance aimed to stitch together from dbt docs and Airflow audit logs and manual metadata. With Foundry's semantic layer, once the device object type is defined with properties from Zabbix, links to alert objects, and links to pipeline objects, every downstream tool, workshop, contour, API, or code is declared and reads from the same canonical definition. I can change the schema once in one place that lives on the object, not scattered across application code.

I am adding features such as Foundry Branching and data connection source agnosticism, particularly the Contour Aggregation Branch. Most people treat Contour as a BI tool, but it is actually more powerful than that. The Aggregation Branch lets me perform multi-step analytical transformations interactively and then publish those as a dataset back into Foundry. An analyst's exploratory work becomes a reusable data asset, not a one-time report. The boundary between BI and data engineering is resolving.

Using Palantir Foundry has positively impacted my organization because I studied Foundry deeply and built the manual equivalents of what Foundry formulizes across connections. I understand what I absolutely need to document, customer outcomes, and what I can totally solve because I know exactly what it costs to solve them.

Palantir Foundry has a steep learning curve and onboarding.

Going deeper, there are additional improvements needed in real engineering solving, such as code reuse and language lock-in. Palantir's lens compute modules were introduced specifically to solve the problem of integrating existing code into Foundry instead of rewriting the logic in Foundry's supported language. I can now containerize code and deploy it directly, with the platform handling scale, authentication, and connection automatically, but this is still maturing. If my team has production-grade Python modules, Kafka consumers, or consumer ML models, I had to prolong rewriting in Foundry's transformation paradigm and maintain a parallel codebase. Neither is acceptable at scale. Improvement is still needed for full parity between containerized workloads and native Foundry transforms in terms of lineage tracking, monitoring, and ontology write-back.

Debugging and observability in pipelines is one of the major drawbacks. However, the platform has many good features and is a good foundation to build upon in the future.

I have been using Palantir Foundry for three years.

Palantir Foundry is a highly scalable solution.

The customer support is great.

I believe the pricing, setup cost, and licensing could be lower.

I have seen money and time saved, but we need to yet see the return on investment.

Before choosing Palantir Foundry, I evaluated other options, including Snowflake and Tableau.

Palantir Foundry's AI capabilities, governance, and security are top tier as it is a fully locked environment and is best for government organizations where data security is a major consideration.

When I use Palantir Foundry's AI features, I find the outputs to be reliable and accurate and have not run into any issues; I trust the results. This is one of the most nuanced topics in the Foundry ecosystem. When the industry pivoted to LLM, the prevailing approach was pointing probabilistic engines at vector databases via standard RAGs, yielding results that were semantically plausible but structurally ungrounded with an unacceptably high hallucination rate and no distinction of facts. Palantir's approach diverges entirely; instead of retrofitting an LLM onto a flat data warehouse, AIP embeds LLM directly into a bidirectional knowledge graph. The resulting architecture dictates that AI interacts with the enterprise through a strictly governed semantic layer that natively understands relationship logic and operational constraints. In plain terms, standard RAG retrieves text chunks and lets the LLM guess connections, whereas OAG retrieves typical ontology objects with live relationships and lets the LLM reason over structured truth. The accuracy difference is architectural, not just prompt engineering.

I would advise others looking into using Palantir Foundry to get in early as soon as you can. If you are mature, do not. I give this review a rating of eight out of ten.

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.