Palantir Foundry Workshop: App-Building Bootcamp

An introduction to both the course, and to Palantir Foundry. An overview and positioning of Foundry. It's an "end-to-end platform" - but what does that really mean? An overview of the sort of clients best suited to using Foundry, key features such as insularity and multi-role collaboration, and an overview of the rest of the course.

A conceptual introduction to Palantir Foundry Workshop with a focus on what it can do, rather than how to use it. Workshop is where data becomes useful to the business. In this lecture, you’ll learn what Workshop is, how it’s used to build operational applications, and how it differs from other apps like Contour, Notepad, and Quiver. This is the layer where users interact with data, make decisions, and run processes.

A guided tour of the Workshop interface. You’ll learn the menus such as File and Help along the top of the screen, and the crucial tabs down the left hand side: Overview, Layout, Variables and Settings. We see how to toggle between Edit and View modes, how to capture which page the user is on in a variable, and how to enable routing (and what that means).

Learn how data becomes available to a Workshop app. We start with a tour of Ontology Manager, where we see the multiple example datasets we will use during the course and see how the datasets link together. We import some datasets into our application as Object sets. We define Object Types and Objects (= Object Instances), which live in the Ontology. And we compare those to Object Set Definitions and Object Sets, which live in Workshop.

Learn how to structure your app using tabs, including enabling and styling a module header, and then making a tabbed section to hold our content. Learn how you could dynamically set the tab, or capture which tab the user is on, via a variable.

You can create any grid-based layout responsively in Workshop if you understand Flex and sections. We add multiple sections, and learn how Workshop decides how much space each takes up. We explore how to avoid common layout frustrations, how to make your app responsive and finally see how to make a fixed-width side bar collapsible.

We add a Filter List widget inside a fixed width collapsible section, powering the widget from properties of an Object Set we created earlier in the course. We add three different types of filters, and see how the default histograms and timelines look pretty but use up a lot of space. We see how to change the default filter visualisation to the more commonly used multi-select dropdown. We learn that one UX goal is to reduce clicks for the user, and this is why radio buttons are preferred for low cardinality properties. And we look at the massive space reduction available with the "pill" layout.

We add multiple sections with a layout type of Row, and compare this to adding multiple sections with a layout type of Flow. We see how the layout panel is hierarchical, and why it is important to keep logically equivalent components at the same level. We see a common mistake in accidentally indenting new sections, and how to avoid this with Flow layout. Finally we see that Flow layout saves time by applying formatting to the parent section, so any new child sections inherit the settings by default.

Explore the remaining - less common - layout approaches for different use cases, including when to break away from standard rows and flows. Specifically, we look at Toolbar, Tabs, Loop and Column layout types.

Create an evenly spaced, responsive KPI panel using the Metric Card widget, by aggregating data into a numeric variable. See how we can style the number using grouping (commas) and change the base type to millions for a better at-a-glance usability. See the different KPI layout options, including card, tag and list and road test the best formatting options.

We can convey a lot more than a single number with a Foundry Workshop KPI. In this video, we learn how to conditionally format our KPI based on thresholds. We also cover event triggering form KPIs and we learn that you can show a spark chart if you add time series objects.

We add a new section and pop a bar chart into it, which is driven by data. We learn how to sort the bars, and learn that you can't filter out the box as you could with Power BI. 

We ramp up our bar chart skills by creating a very pretty stacked bar chart, complete with legend and user-optimised sorting and labelling.

Foundry has a vast range of variable types, and it can be really daunting to get started. We run through the main variable types, demystifying them with clear examples. Specifically, we look at the first menu, which defines the output data type (String, Boolean, Numeric, Date) and see the secondary menu type (Static, Function, Object Set Aggregation, Object Property etc.) defines how the output is calculated. We see that the secondary menu repeats almost exactly for most of the data types.

We link our filters panel to our data through the use of variables. We assign a variable to hold the choices made in the filter panel, and use that variable to drive what shows in our visualisations.

Until this point, we have used simple pre-defined aggregations of data in our KPIs (sum, count, avg). Now we want to create proportion of delayed flights, and for that we want to exclude 23k nulls and negative delays (where the flight left early). We explore the no code option for doing this, creating several variables and dividing some by others. We learn it is best to be rigorous in variable naming to create a manageable outcome and see that it would be quicker to use a function (repo) where possible.

Learn how to show or hide components based on variable values. This is a key step in creating guided, user-friendly apps.

Build more advanced variables that combine logic, conditions, and transformations. Starts with variable transformations, and shows how to replace a variable that depends on other variables, with one that contains all its own logic. Then briefly looks at Struct variables and Arrays.

We create two widgets: Object List and Property List, and see how to link them to each other, so that clicking on the high level Object reveals whichever details you choose in the Property list. We consider the risks involved in this, by using Aircraft data. Each aircraft has a unique serial number but potentially one or many Tail Numbers (like registration plates). We use this one-to-many relationship to explore how an Object-property list could be misleading.

Learn how to use Foundry pivot tables to group and aggregate data, transforming detail into meaningful summaries that power analysis and decision-making.

Learn how Foundry’s operational and data quality widgets make data health visible, helping users quickly spot problems before they affect downstream analysis. See the benefits of exposing this to your users, rather than hiding it away. We cover Data Freshness and several other widgets that lend themselves to operational data.

Unlike Power BI, Foundry enables two-way interaction with data, allowing users to write changes back to the database. These updates are not ad hoc — each possible change must be implemented as a predefined action, giving you fine-grained control over what users can modify and how those changes are applied.

In this conclusion, we recap the key concepts covered in the course and outline how you can continue building confident, data-driven applications in Foundry Workshop.