News & Intel
LiveDaily AI-curated intelligence on Palantir Foundry, Ontology, AIP, Apollo, contracts, and community feedback. Updated automatically via GitHub Actions every day at 7 AM UTC.
Limit to only 10 objects Object table
Is it possible to direclty only include only 10 objects in a workshop object table widget? 2 posts - 2 participants Read full topic
Help with use case: Exporting a sorted excel file from workshop / sdk for an object type ~100k
Hello, is there any way I can export 100k rows of an object sorted from workshop, pipeline builder or a custom code repository? My client’s use case involves capturing snapshots of a large dataset (preferable an excel file) that is sorted by multiple columns from our main workshop app. I tried building a custom CSV file in code repositories and sorting the dataset there, but for objects more than 100k the function times out. Excel type exports from the export feature in workshop buttons enable objects more than 100k, but there is no built in sorting feature. Any guidance would be appreciated, thanks! 1 post - 1 participant Read full topic
We translated the Palantir manifesto for actual human beings
HN Discussion — 1 comments · 5 points.
Need access to Foundry
Hi, I’m currently taking the “Speedrun Your First AIP Workflow” course on Palantir Learn. I’ve completed the setup steps but don’t have access to a Foundry instance. Is there a way to request access to a training or sandbox environment to follow along hands-on? Thanks in advance! 3 posts - 3 participants Read full topic
Best approach for incremental processing when S3 source provides full snapshots
We’re ingesting data from an S3 bucket where an upstream Spark job writes a full snapshot (single file, fully replaced) on a scheduled basis. Each snapshot contains all rows — both unchanged and new/modified records. Our data has a unique key and a last-modified timestamp column. Our challenge is that we don’t want to process the entire snapshot through our downstream Pipeline Builder pipelines every time, as they include compute-heavy transforms and LLM calls. Our current approach is to use a Python transform to compare each new snapshot against the previous state, detect only the new and modified rows (delta), and append only the delta to a master dataset. This allows downstream Pipeline Builder pipelines to read the master dataset with the incremental input toggle enabled, processing only the newly appended rows. The Pipeline Builder output is then set to Snapshot Replace for deduplication. However, since the Snapshot Replace output is a replace transaction, any further downstream
Zero-shot TimesFM on Foundry: Model Adapter + Live Inference pattern — and three questions
Shipped this last week and figured I’d share the shape, because honestly I couldn’t find anyone on here who had done TimesFM on Foundry yet and I’d rather compare notes than keep guessing alone. If someone has a cleaner version of this I’d genuinely like to hear it. One note before the code — yes, Model Studio already has a native time-series forecasting trainer (AutoGluon under the hood, ensembles AutoETS, DeepAR, PatchTST, TFT, TiDE, Theta). If you have labeled history per series, start there. This post is for the neighbouring case: cold-start series, brand-new sources, no labeled history yet — where supervised per-series training doesn’t really apply and a zero-shot foundation model can carry you until you’ve accumulated enough training data to go back to Model Studio. The shape, short version: checkpoint as a Foundry dataset → Model Adapter → Live Deployment → named Function → Workshop widget. No fine-tuning. Bence TimesFM is good enough zero-shot for daily-granularity foreca
How can I call Foundry APIs from Foundry itself?
For example, if I want to leverage ontology APIs, grab build history, or update users programmatically? 2 posts - 1 participant Read full topic
Feature Request: Programmatic Webhook Management in Data Connection
Summary Data Connection webhooks currently require 100% manual configuration through the UI — there is no API, CLI, SDK, or automation support for creating, reading, updating, or copying webhooks between sources. This makes webhook management highly “click-ops” and creates significant friction for teams managing webhooks across environments (dev → staging → prod). Problem Statement Webhooks are a critical integration capability , enabling Foundry applications to interact with external systems in real time. However, the current workflow for managing them has several pain points: No programmatic access — There is no API to read or write webhook configurations. Every webhook must be manually created and configured through the Data Connection UI. No environment promotion path — When moving from dev to prod, teams must manually recreate each webhook on the production source by hand, copying over the HTTP method, path, headers, query parameters, body templates,
Standard Chart filter interaction
Hello! I’m new to the workshop and still need to learn a lot! I’m seeking for a help or assistance. In the workshop, I created a filter list. Using the same object set and the filter output as a selection, I created both a bar chart and a pie chart. However, they are grouped by different categories. Is there a way to make these two charts interact with each other—similar to a contour analysis—so that selecting or drilling down in one chart affects the other? 1 post - 1 participant Read full topic
pltr-cli 0.15.0
Command-line interface for Palantir Foundry APIs
Dataset JSON Parse
I have a dataset where I am appending ndjson files, and it’s forcing me to use csv parser when I auto infer the schema. I tried forcing the schema to use a json parser and get com.palantir.foundry.spark.api.errors.UnsupportedFoundrySchemaException Please help thank you 1 post - 1 participant Read full topic
PB Feature Request: Get dataframe().columns
In code repo, I use input_dataset.dataframe().columns to get all the columns. For example: string_columns: list[Column] = [F.col(c).cast(T.StringType()) for c in df.columns] df: DataFrame = df.withColumn('check_changes', F.concat_ws('_', *string_columns)) In Pipeline Builder, I need to carefully click “Add Item” for all ~100 columns, in the correct order, in the concatenate string transform. Please enable an option to either select all, or preferably a way to paste a list of columns, similar to how the Select transform allows for pasting a list of columns. (This would also help with the pain of manually dragging the handles to re-order 100 columns.) 4 posts - 2 participants Read full topic
AI Native Market Research and Forecast Report 2026-2032 with Deep Dives Into Cursor/Anysphere, Lovable, Replit, Tessl, Windsurf, and the Hyperscale Plays from Microsoft, Google, and AWS
The main market opportunities are in AI-native development platforms transforming software engineering. Organizations are urged to innovate with AI-augmented systems, autonomous agents, and bespoke applications. Key areas include hyper-agile engineering, AI-d…
Unable to delete Marketplace store in Foundry Developer Tier
Hi, I am using Palantir Foundry Developer Tier and I created a Marketplace store. I am unable to find any option to delete the store from the UI. I wanted to confirm: Is it possible to delete a Marketplace store in Developer Tier? Or is this restricted to internal/admin-level operations? I have already removed all products from the store. Any clarification would be helpful. Thanks. 1 post - 1 participant Read full topic
I am unable to create or add users as the option is not available for my account, and I also do not have permission to delete stores from the marketplace in Palantir Foundry.
I am unable to create or add users as the option is not available for my account, and I also do not have permission to delete stores from the marketplace in Palantir Foundry. 1 post - 1 participant Read full topic
Feature Request: Add Data Expectations to Pipeline Builder DSL
Summary Pipeline Builder supports configuring data expectations (Primary Key, Row Count, Value is not null, etc.) through the UI, but these are not exposed in the DSL . I’d like to request that expectations be added as a first-class concept in the Pipeline Builder DSL so they can be managed programmatically. Motivation When building production-grade pipelines, data expectations are essential guardrails. Currently, if you author or modify a Pipeline Builder pipeline via the DSL (e.g., through AIP / AI-assisted development, or programmatic pipeline management), there is no way to define expectations in code . They must be manually configured through the UI after the fact. This creates several problems: Expectations are invisible in the DSL — When reading a pipeline’s definition via get_pipeline_builder_definition , expectations are not included. There’s no way to audit or review them programmatically. AI-assisted pipeline development can’t add expectation
Feature Request: Allow AI FDE to Submit Internal Product Feedback
Feature Request: Allow AI FDE to Submit Internal Product Feedback Summary When AI FDE encounters platform limitations, missing APIs, or has to fall back to manual workarounds, it should have the ability to automatically file structured product feedback tickets to Palantir’s internal teams. This would create a continuous, high-quality feedback loop between real user workflows and platform development. The Problem Today AI FDE works across thousands of user sessions, building pipelines, writing functions, configuring ontologies, and more. When it hits a limitation — like a missing DSL feature, an unsupported tool capability, or an API gap — the only options are: Tell the user to do it manually (breaks the automation promise) The user files a forum post (requires effort, often doesn’t happen) The friction is silently absorbed (Palantir never hears about it) Most of the time, option 3 wins. The feedback is lost. The Opportunity AI FDE is arguably the
Package Object instances + their edits in a marketplace product
I would like to package an object type, with its object instances , created via manual Actions execution, into my marketplace product. Is it possible ? What is the easiest way to package those edits ? (assuming this is a backing-dataset-less object type, where for example most of the properties are edits-only) 1 post - 1 participant Read full topic