Palantir Foundry, a solution serving Data

Context

What do the CIA, Scuderia Ferrari F1, Airbus, Sanofi, Fiat Chrysler and Credit Suisse have in common?

At first glance, it is difficult to find an obvious link between the famous American intelligence agency, a Formula 1 team, an aircraft manufacturer and a pharmaceutical company. But the answer lies in the use of Palantir Foundry ; this digital solution developed by Palantir technologies has become more and more essential in the world of data. At least, this is the ambition of its CEO Alex Karp who declared a few years ago that “Palantir was founded to create software for the most important institutions in the world”.

In this article, let's try to explain what need the Foundry platform addresses. How does she do it? What are the use cases? Finally, let's try to explain why we decided, at Aqsone, to develop expertise in Foundry.

What needs does Foundry meet?

Many businesses generate an astronomical amount of data during their daily operations. Think of the data generated by an airplane during a flight, or the data generated by car part production machines on an assembly line. These companies often operate in separate silos. For example, production data will be found in systems different from those for maintenance. Very little, if any, visibility is shared between these two professions, yet potential for cooperation could exist if the two professions understood each other. You can imagine the complexity that can exist between the different silos of industrial companies with tens of thousands of employees, hundreds of professions and dozens of different departments.

Having data distributed across several non-communicating silos is one of the first obstacles to digital transformation. This creates communication barriers between employees and prevents the creation of a climate conducive to synergies, innovations and improvements that can emerge between the company's different professions.

The integration of data into a single platform (often called data lake) is necessary to enable the proliferation of data-driven practices and digital-based services. It's here that Palantir Foundry intervenes brilliantly.

How Foundry addresses the need for data integration?

To reduce data silos, Palantir developed Foundry. This platform allows you to:

• Integrate / extract / collect data in a single secure, cloud-based platform (with the possibility of using custom storage). Data ingestion is possible from the company's various systems (ERP, MRP, IoT, etc.);
• Store and manage data: access management for authorized users, structure/classify data into objects easily recognizable by all business areas of the company (data objects/ontology), automatic updates of data to ensure their reliability ( scheduling), data quality monitoring (data health);
• Manipulate data: pipeline development to collect, transform and store data in the format that meets the needs of each team;
• Analyze data: simple and quick interactions with data, creation of dashboards, speed of analyzes carried out on massive data and use of numerous Data Science or Artificial Intelligence tools.

Thus, these capabilities allow engineers, data scientists or non-specialists to intuitively access the most relevant information in order to improve their jobs.

What use cases for Foundry?

Reduce delivery delays on aircraft:

Airbus has used Foundry since 2017. The platform allows it to collect, extract and centralize the set of heterogeneous data necessary for the production of the A350, including that coming from suppliers and that installed on the tablets of engineers and operators working on the A350. All assembly, operating and safety data is consolidated on Foundry and presented in the form of configurable dashboards. This made it possible to better support the ramp-up in A350 production, and also to better anticipate and react to non-quality problems, likely to delay delivery of the aircraft.

Find the best technical configuration for a Formula 1 car:

Foundry allows Scuderia Ferrari to combine the technical expertise of the team's engineers with the immense amount of data at their disposal. Telemetry, spare parts, simulation, testing and pilot feedback data are integrated within Foundry. Instead of spending their valuable time integrating and cleaning this data, Ferrari engineers in the factory or on the track can focus on their most important tasks, using the data to make faster, better-informed decisions. Foundry made it possible to have a digital twin of the car, on which engineers rely in order to better analyze performance, adjust their hypotheses and iterate quickly in order to converge on the best technical configuration to use for a given circuit .

Why is Foundry so successful?

Data-centric: 

Unlike Microsoft Azure, AWS or Google Cloud Platform which represent very general cloud services,  Palantir Foundry is data-centric: the platform is optimized for data ingestion, data engineering, data visualization and machine learning.

Ease : 

Adopting and using Foundry is very simple. On your internet browser, you will have all the tools pre-installed, pre-configured and maintained by Palantir. A one-hour training session will be enough for any operator to create filters, aggregations and simple graphs that meet their use case. Foundry contains both code-based tools (code repositories & code workbook) for experts, or non-code based (outline & data preparation) tools for non-technical beginners. This makes data manipulation accessible regardless of your level of technical expertise and gradually removes the psychological barriers that one may encounter for the adoption of data-driven practices.

Versatility:

Foundry is designed to handle massive data, structured or unstructured (including time series). It allows data ingestion, synchronization and monitoring. Pipelines can be developed collaboratively (Git) on several computer languages (SQL, Python or Java). Their update can be scheduled. A data lineage functionality allows the pipeline to be displayed on a user-friendly graphical interface (monocle) much appreciated by data engineers. THE data analysts can create their data visualization using code-based tools for the most complicated visualizations or non-code-based tools for the most common visualizations (bar chart, histogram, pie chart etc.). THE data scientists will find the necessary tools to help them from developing a machine learning prototype to putting it into production.

Why is it important for Aqsone to master Foundry?

Thanks to its many qualities, the Foundry platform can become essential for successful digital transformation. “We allow companies to regain power over the gold mines that are their data, and thus be able to catch up with digital-native companies,” assures Fabrice Brégier, the boss of Palantir France, the former CEO of Airbus.

We cannot ignore the success of Skywise, the Airbus Palantir Foundry, which in just a few years has become the star of the aeronautical world by integrating not only the manufacturer's data but also the data of around a hundred airlines and suppliers; the ambition being to become the beating heart of aviation. This is also the slogan of Skywise. 

Aware of the importance that Foundry will have among our clients in the coming years, Aqson has developed expertise on this platform and intends to rely on it in order to bring more value to its clients during their transformation and data science projects. From ingestion to operation, including machine learning, we support our customers from their first steps on Foundry and we offer them use cases inspired by our experience.