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.
Pilot - Error response
Hi, I am trying to use Pilot to build out my front end app. I already made the ontology. I am able to load into the application, and the LLM processes my request, however it doesn’t actually make any edits to the code even though it says it’s done the front end. I then ask it to add the code but i keep getting the message “Error communication with language model service” Any help with this? Kind regards Sam 2 posts - 2 participants Read full topic
Filter on Array Column via Typescript OSDK
Looks like in Typescript V2 repos I can’t filter on Array Columns? Or am I just not using the OSDK correctly. At least I couldn’t find the right API syntax in the docs. https://palantir.mcloud.merckgroup.com/docs/foundry/ontology-sdk/typescript-osdk/ E.g I was trying: orders.where({ manufacturNames: { $eq: “name“ } }) manufacturNames is an Array String Field. I think in the Typescript V1 Repo’s this kind of filters are possible as well as in Workshop. 2 posts - 1 participant Read full topic
Timer Tool, wait for backend things to be available
Might be a stupid suggestion but after creation of ontology objects need to wait for the them to index, I would like to be able to call a timer, have AI FDE wait a couple of minutes and then create typescript functions. (Or is there something else I can ask it to accomplish this?) 1 post - 1 participant Read full topic
AIP Logic can't use Ontology types backed by Restricted Views with Project-scoped execution?
I am trying to create an AIP Logic to create an AI Agent that uses the Ontology SQL tool. It currently uses an object that is backed by a Restricted View and we were using the User-scoped execution that comes by default. Everything works great. However, it’s important that other users also have access to the logs, so we would like to change it to Project-scoped execution. But, when we try to do this we get errors: If you try to Preview it shows: “Publish errors: Unable to create scoped token” If you try to Publish the function it shows: Failed to publish: {"type":"STATUS","status":500,"body":{"errorCode":"INTERNAL","errorName":"Default:Internal","errorInstanceId":"a6d86274-1b0b-4597-a92e-7a5a7957a31f","parameters":{}}} I tried to add the Restricted Views as File References in the AIP Logic Project, but it does not work. Even trying to create the AIP Logic in the same Project as the Ontology, it throws the same error. Any ideas on how to solve this? 1 p
Batched Execution Function-Backed Action Logging
I’ve implemented a create action for an ontology object where a user uploads a csv file and the action processes all the rows in batches. This was because without batching I would get an error about having a limit of 20 rows. However, now I am not able to enable action logs. Any suggestions on how best to approach this? I need the logs for audit purposes. 1 post - 1 participant Read full topic
Map Widget in Ontology
Hello! I have a dataset deployed to ontology with a geopoint column cast to geopoint type. In Workshop map widget I selected the object layer and set the property to geopoint but no points are showing on the map. What am I missing? 1 post - 1 participant Read full topic
Ontology architecture in Workshop
Hi everyone, We are currently working on designing an ontology architecture in Workshop and would like to get some input and recommendations from the community. Right now, we are debating whether it is better to: use one large denormalized (flattened) object, or keep multiple normalized linked objects that reflect the real relationships in the ontology. Key questions Performance: normalized vs denormalized From our experience, working with many small objects with multiple links and transversals seems to noticeably worsen Workshop performance, especially in dashboards and derived properties. Because of this, we are leaning towards using a single large denormalized object instead of multiple linked ontology objects. Is this a common experience, and is it generally recommended to move towards a more flattened object structure in Workshop to improve performance? If so, what is the practical benefit of maintaining a fully normalized ontology with many linked objects
Auth failing for Anthropic Agents SDK and LLM Proxy
When building with Anthropic’s Agent SDK (not the client SDK, that works fine) I set the ANTHROPIC_API_KEY and ANTHROPIC_BASE_URL to my Foundry URL and access token. I can see by proxying the requests API calls to my Foundry stack and to Anthropics APIs, for example: [REQUEST URL] CONNECT <REDACTED>.palantirfoundry.com:443 HTTP/1.1 [HEADER] host: <REDACTED>.palantirfoundry.com [HEADER] connection: close [REQUEST URL] CONNECT api.anthropic.com:443 HTTP/1.1 [HEADER] host: api.anthropic.com:443 [HEADER] proxy-connection: close The fill stack URL provided in the is: <REDACTED>.palantirfoundry.com/api/v2/llm/proxy/anthropic The token is a valid token create with the OSDK. I also tried a personal access token. But I keep gettin 401 authentication failed. attempt = 1 error = 'authentication_failed' error_status = 401 max_retries = 10 retry_delay_ms = 536.6033701315846 session_id = <REDACTED> subtype = 'api_retry' type = 'system' uuid = <REDACTED&g
Pilot Feedback - Rate Limits
Sharing some feedback for Pilot on Dev Tier I was able to rebuild an app I’d previously built in Developer Console w/ VSCode and Continue with three prompts and 25 minutes, pretty impressive The one thing having reviewed the log files for the build is that at the 130k tpm rate limit in place for this tool on Dev(I assume it’s higher on Enterprise) >80% of calls failed on my first build. It’s totally understandable why limits exist for us free users, however the limits are quite impactful, the build time would be significantly faster with a 225k tpm rate limit. Agent Successful Failed (Rate Limited) Total Attempts Failure Rate Avg Tokens/Call Peak Tokens/Call ontology-architect 3 14 17 82.4% 27,582 35,522 designer 1 4 5 80.0% 3,234 3,234 frontend-architect 17 71 88 80.7% 38,133 52,662 seed-data-generator 1 5 6 83.3% 26,304 26,304 main-orchestrator 4 16 20 80.0% 25,2
Pilot should have a "no ontology modifications" mode
I’ve been testing Pilot recently. Rather than use the existing properties on the ontology objects provided as inputs, it will often make up new properties that it thinks better suit the app you’re trying to build. This is quite an insideous problem as unless you know the ontology intimately, these can look like valid properties and so get merged in as changes along with other valid modifications. This creates duplicated and “dead” properties. Can there be an option to limit Pilot to building using only existing object properties, links and action types? This would make it much safer to iterate on modules when an incumbent ontology already exists. 1 post - 1 participant Read full topic
Ontology as Code API Name Validation Failing
I am getting errors related to my API names that I suspect are due to length: { "type": "typedBlockInstallServiceValidationError", "typedBlockInstallServiceValidationError": { "blockType": "ONTOLOGY", "error": { "severity": "BLOCKING", "error": { "type": "linkTypeInvalidApiName", "linkTypeInvalidApiName": {} }, "fallbackMessage": "The specified API Name for the LinkType is invalid. See documentation for ApiName for details. [linkTypeRid=ri.ontology.main.relation.1772999f-f3d3-4759-b943-6836d1a108dc, apiName=investmentGroupOwnershipsinvestmentGroupsToInvestmentGroupOwnershipsByParentInvestmentGroupsPkLinkToMany, linkTypeId=c2udmhwk.investment-groups-to-investment-group-ownerships-by-parent-investment-groups-pk-link-one]" }, "traceId": "8974edca34d93f18" } } { "type": "typedBlockInstallServiceValidationError", "typedBlockInstallServiceValidationError": { "blockType": "ONTOLOGY", "error": { "severity": "BL
Issue with Pilot
Hey folks I saw the announcement about Pilot , and tried it out but I keep getting the same error when I hit “starting servers” after adding a prompt, ontology and hitting go. same issue across multiple browsers/OS - am I doing something wrong? 1 post - 1 participant Read full topic
Geopoint Object Property Compatibility Issue
Hi, I have made a pipeline builder that creates a geopoint. When I preview in pipeline builder, it shows it as this type. (Picture Below). Once the dataset is built it shows it as Struct, but of typeclass: geoPoint.geoPoint (Picture below) However, when I go to create my ontology object, it sets the property as ‘Struct’ and I can’t choose geopoint. What am I doing wrong? Kind Regards Sam 4 posts - 3 participants Read full topic
Do ontology objects created with the OSDK/Maker SDK support groups?
The define object type definition does not appear to support groups: type ObjectTypeDefinition = { apiName: string; primaryKeyPropertyApiName: string; displayName: string; pluralDisplayName: string; titlePropertyApiName: string; properties?: { [key: string]: ObjectPropertyTypeUserDefinition; }; implementsInterfaces?: Array<InterfaceImplementation>; description?: string; icon?: { locator: BlueprintIcon; color: string; }; visibility?: Visibility; editsEnabled?: boolean; status?: ObjectTypeStatus; datasources?: Array<ObjectTypeDatasourceDefinition>; aliases?: Array<string>; }; I also do not see a way to create groups. I assume this is a roadmap item, but if there is a way that I missed please LMK. 1 post - 1 participant Read full topic
How to Trigger Ontology Action Type
Hi Experts, Currently, I need to set up the rules to trigger action in one business scenario,like the machine’s temperature exceeds 120 degrees, triggering an alarm notification(Email or SMS). And I have build the ontology already. I wonder how to set rules to trigger ontology action in Foundry. I have been reading some palantir offical documents, like Action Rules/Foundry Rules. But I don’t find the right solution. Could you please guide me how to set up this? Thanks in adavance! Peton 1 post - 1 participant Read full topic
AI-FDE Core Architecture Library
Palantir explicitly structures AI FDE around two tiers: Modes (the broad task at hand, e.g., data integration or ontology editing) and Skills (granular capabilities usable across multiple modes). This GitHub repository library mirrors this exact two-tier hierarchy — skill.md maps to Palantir’s Skills layer, agents.md maps to full Mode-level agent configurations. Your prompt library will sit as the execution layer beneath both. https://github.com/s-andthat/palantir-ai-fde-library Hopefully someone will find this useful. 1 post - 1 participant Read full topic
Linked objects and action types
Hi, In my ontology manager on object type A, I have actions to link with other object type B. When I click on link button in main branch, A gets linked to B. But the same action throws an error in a branch. Why? I tried indexing object types A and B in my branch. It still does not work. Thank you, 2 posts - 2 participants Read full topic
RemoteException: INTERNAL (Default:Internal) when writing embeddings to Ontology object via AIP Logic Action
Problem I’m building an AIP Logic pipeline that processes a PDF and creates a Knowledge Article object. The workflow is: Extract text from a PDF media reference Use an LLM to generate structured fields (Title, Short Description, Content, Category, Tags, Key Points, Content Type, Entities) Generate text embeddings using text-embedding-ada-002 Convert the embedding output to Array Call a Create Knowledge Articles ontology action to write the data to an object All upstream blocks execute successfully during preview (PDF extraction, LLM output, embedding generation). However, the final Ontology Action fails. Error RemoteException: INTERNAL (Default:Internal) Trace ID: 11896d79c7214279 Error ID: b4d93c5-95df-47fb-a560-f765603daa8b The failure occurs specifically in the Create Knowledge Articles action block. In the debugger, the action shows: Arguments: {} Even though all fields appear correctly mapped in