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.
Trump’s Investments In Palantir And Nvidia Draw Scrutiny—Here’s What Companies Were Traded
The president made between $220 million and $750 million worth of securities trades in the first quarter of this year, according to a Reuters analysis.
Daily Hacker News for 2026-05-15
The 10 highest-rated articles on Hacker News on May 15, 2026 which have not appeared on any previous Hacker News Daily are: <ul> <li> New Nginx Exploit (comments) </li> <li> First public macOS kernel memory corruption exploit on Apple M5 (comments) </li> <li>…
If This Is the End Times, I’ll Be Your Huckleberry: Peter Thiel’s Antichrist Theology
It’s May 15th, 2026 — recording live from what Bo accurately described as Jurassic Park, because Raven the chicken was laying an egg on my thigh while we were discussing the Antichrist, and Marceline was on my back pecking my head before we hit record, and I …
Institutional investors flocked to establish new stakes in semiconductor firms in first quarter
Institutional investors actively bought shares in semiconductor companies such as Micron and Intel during the first quarter. This move positioned them to benefit from a significant market rally. Demand for memory chips, fueled by AI development, drove substan…
Palantir's SaaS is dead claim is a warning shot for founders
HN Discussion — 1 comments · 2 points.
Osiris – An open-source Palantir alternative for global intelligence
HN Discussion — 0 comments · 5 points.
Palantir — не дата-компанія («Пояснюємо Palantir», №1)
Опубліковано: 11 листопада 2020 р. Оригінал англійською мовою: https://www.palantir.com/palantir-is-not-a-data-company/ Примітка редактора: це перший допис у серії « Пояснюємо Palantir », у якій ми розглядаємо широке коло тем — від нашого підходу до приватності та безпеки до питань безпечного застосування AI/ML тощо. Palantir часто називають закритою компанією. У цьому є частка правди. Упродовж багатьох років ми працювали переважно з інституціями, що мають виняткові вимоги до конфіденційності, — у сферах оборони та розвідки. Нерідко ми просто не мали вибору, окрім як зберігати мовчання щодо своєї роботи, навіть тоді, коли в медіа чи публічному просторі з’являлися хибні уявлення про природу нашого бізнесу. Сьогодні, коли ми працюємо з клієнтами у значно ширшому колі галузей, у нас з’явилася можливість бути більш відкритими. Це особливо актуально для секторів на кшталт охорони здоров’я, де програмне забезпечення Palantir використовується для обробки персональних даних. Люди ма
Palantir, як і раніше, не є дата-компанією («Пояснюємо Palantir», №7)
Опубліковано: 6 червня 2025 р. Оригінал англійською мовою: https://www.palantir.com/palantir-is-still-not-a-data-company/ Примітка редактора: це сьомий допис у серії «Пояснюємо Palantir» ,, у якій ми розглядаємо широке коло тем — від нашого підходу до приватності та безпеки до питань безпечного застосування AI/ML тощо. У цьому дописі ми повертаємося до деяких тем, обговорених у нашому першому матеріалі 2020 року, та відповідаємо на типові непорозуміння стосовно того, ким ми є і чим займаємось. Попри численні масштабні зрушення, зміни та події останніх років, одне залишається незмінним: Palantir, як і раніше, не є дата-компанією. Хто ми і чим займаємось Palantir створює програмне забезпечення, яке організації використовують для того, щоб краще керувати своїми даними, удосконалювати операційні процеси та якісніше обслуговувати людей, які на них покладаються. Ми пишаємося тим, що критично важливі організації — серед них ті, що рятують життя, поліпшують клінічні наслідки, в
Unable to Reference Newly Created Object in 'Create Link' Rule in Ontology Manager Action Type
I’m having a issue with an action form. I am configuring an action type in Ontology Manager to allow users to create a shipment and link it to another object type called Project via a dropdown. Both object types have already been created and populated as well as their link type via a many-to-many join table (so not a direct property like a foreign key). I set up a “Create object” rule for Shipment and then a “Create link” rule following that for the Shipment-to-Project link. But the issue is that in the “Create link” rule configuration, there is no UI option to reference the Shipment object just created in the “Create object” rule as the source for the link. The UI only allows me to select parameters exposed to the form, not the output of the previous rule. As a result, I cannot configure the action to create and link the Shipment in a single form submission without requiring the user to select a Shipment (which does not exist yet). Am I missing something here? Is there a way
Feature: Branching Support for Foundry Rules Engine / Rules Widgets
As far as I am aware there is no clean way to test rule changes without affecting downstream datasets as the workshop widgets don’t support branching. I understand I can use the rule preview widget to “faux-test” but you push it downstream without building the dataset. It would be nice to have branching support for these widgets/app 1 post - 1 participant Read full topic
Automation debugging and understanding behavior
Hello Community, I have an automation setup and have a few doubts regarding failure handling and debugging. Scenario My automation takes in an object set, applies a filter, and then the filtered objects trigger the automation. The automation then runs a function to update another object set. For example: Object Set A contains a large list of drivers across multiple regions along with their license numbers. Since the dataset is very large, I created multiple automations, each filtering drivers for a smaller region. These filtered objects trigger a function that updates Object Set B, where driver details and license numbers (as foreign key) need to be populated. Doubt 1: Behavior when automation fails Suppose the automation is triggered by 100 objects. The function processes the first 70 objects successfully. At the 71st object, there is an issue (for example, a validation condition causes the loop to get stuck and the function eventually
Mounting multiples files for docker container
Greetings, I’m deploying model objective via docker images. Dockerfile is like this FROM <MODEL_BASE_IMG> USER root WORKDIR /models RUN getent group 1000 ... COPY /folder/model_checkpoint /models/checkpoint RUN chown -R 1000:1000 /models USER 1000 EXPOSE 8080 ENTRYPOINT [...] CMD ["/models/checkpoint", \ "--host", "0.0.0.0", \ "--port", "8080", \ ] Each time I update the model with new model_checkpoint file, I have to build the image again while there’s no change in the environment set up. It’s quite time-wasting when the model_checkpoint file is GBs large. Is there any way I separate the model_checkpoint file and the Docker image? Like storing model_checkpoint file in some dataset, and when doing the docker run, it gets the latest model_checkpoint file from that dataset to run, then I don’t have to build the image again. Thanks in advance. 1 post - 1 participant Read full topic
Expose AIP-Analysis Functionality in OSDK
Hello, AIP-Analyst works surprising well & finding that users respond very well to the analysis it can generate. I’m noticing more users asking for custom workflows on top of AIP-Analyst (more curated front-ends, more restricted set of objects types it is able to work with, etc.). While the workshop AIP-Analyst widget works great within workshop, we are seeing more requests for this type of functionality with OSDK applications. Some specific things I’d be trying to accomplish: Pre-aggregate some object data (according to the businesses logic) to pass into AIP Analyst Custom configs (units and/or colors) for any graphs/charts it generates Specific language / verbiage the output should be using when communicating. I’m sure a lot of this can be configured with a detailed prompt into vanilla AIP-Analyst, but it would be nice to preconfigure this for the user such that burden isn’t on them to get the analysis they want. Thanks! 3 posts - 3 participants
Building agentically with AIP
Hey! It’s been a few months since I last used Foundry, and I’m noticing lots more tools available. Like new generations of AIFDE, Pilot, improved MCP, etc. I’m wondering if there is a single point of reference about how they all link together so we can best make use of them. I’m particularly interested in agentic workflows working symbiotically with agents such as Claude Code // Codex. Moreover, what is Pilot and how should we think about it vs AI FDE and productisation? Thanks! 1 post - 1 participant Read full topic
On Interpolatable Archives
AI Latent Spaces as Shapeshifting Skeleton Libraries and Explosion Drawings bearing cognitive hazards and new opportunities to play.
A European central bank has signed a mega deal with a cloud service provider. The problem for Google, Microsoft and Amazon? It’s not with them
Originally built to support the retail business, Schwartz Digits is now a trusted provider of secure data services to European businesses and governments.
The potential new Prime Ministers, ranked from most to least terrible
The potential Labour leadership challenge offers a rogue’s gallery of smarmy weasels, shameless sellouts, ineffectual has-beens, and someone who is kind of OK
The potential new Labour leaders, ranked from most to least terrible
The potential leadership challenge offers a rogue’s gallery of smarmy weasels, shameless sellouts, ineffectual has-beens, and someone is who kind of OK