Product User Community

Does Cognite AI search multiple documents when you ask a question, or does it stop as soon as it finds an answer?When I selected multiple documents in Canvas and asked AI a question about them, it seemed to only give me answers from one of the documents. 

I have changed my mobile phone as last one is lost. Trying to set up microsoft authenticator in order to use cognite tool but not able to set authenticator. Please help on this.

Hi Everyone,I’m reaching out to the community to gather valuable feedback and insights on Cognite Data Fusion (CDF), specifically its Canvas + Charts features as applied to Root Cause Analysis (RCA).How are these tools impacting the RCA process? In your experience, how have the quality, dependability, and response time of RCA processes improved since implementing CDF? What kind of improvements have you observed in identifying root causes, especially in terms of speed and accuracy? With the inclusion of Generative AI (GenAI) within CDF, how has AI-driven data exploration changed your RCA workflows?I’d greatly appreciate hearing about how the combination of Canvas, Charts, and GenAI is transforming the way you approach RCA, and any real-world examples or challenges you’ve faced.Looking forward to your insights!RegardsMuhammad Waqas 

I’m just discovering the existence of the Admin Workspace in exploring about location filters. I do see the Location filter in the “Add Data” UI in Charts. I do not see evidence of the Admin Workspace in my CDF UI so that I can define Locations. Does this need to be enabled first? 

A chart in Canvas should update if it is changed in Charts. For example, I added a Chart to my Canvas. Then I added another timeseries to my Chart. But in Canvas, the chart didn’t update with the new timeseries. 

Hi there,Before I begin, some terminology to make sure we are on the same page.CDF tenant = CDF project (these have been used interchangeably over time, not sure which is applicable at the moment, I will be referring to it as tenant)CDF cluster = The cloud cluster where the CDF tenants/projects/instances are hostedNow, let’s begin.At Aker BP we have been migrating over to a dedicated CDF cluster on Azure and we have been experiencing some pain points with regards to the compute resources being shared for all tenants on the cluster. Previously our CDF tenants were hosted on GCP on a cluster shared with other companies, this cluster is scaled to handle multiple customers and so the base scaling of the compute resources is much higher than it would be on a dedicated cluster, and as such these problems are less likely to be encountered, not impossible just not as likely. The problem we have seen is that if we from our dev tenant overload the API then the API becomes unavailable for all our

Hi everyone,I’m exploring the possibility of using the Shap library for model interpretability in Cognite Jupyter Notebooks, but I understand that Pyodide has limitations with packages requiring native libraries. According to the Pyodide FAQ, such packages may not be supported.Would it still be feasible to include Shap, or is this limitation a blocker? If it’s not possible, are there any recommended workarounds, such as using a different package?Thanks for any guidance or insights!Best regards,

Hi! I’m working on a Equinor project where we’re using the Cognite GraphQL. When using the list-query with the sort property it returns more items in the result than it does when there sort property is not included. The only difference in the query is the sort-property. Our dataset has more than 1000 items, so we’re fetching paginated data. I’ve been told by a team member from Cognite that this is a known bug, however, we would very much like to know what the timeline is for fixing it? We need the sort functionality so if it is not being prioritised we will need to implement a workaround. But would of course prefer not to have to do that. Thanks!

We are trying to ingest the data from csv file into Staging table. Below is the config.yaml file.logger:  console:    level: INFO# Information about CDF projectcognite:  # Read these from environment variables  host: https://fusion.cognite.com/  project: sandbox  idp-authentication:    # OIDC client ID    client-id: abc345    # URL to fetch OIDC tokens from    token-url: https://login.microsoftonline.com/gvbiy76/oauth2/v2.0/token    # Alternatively, you can specify an Azure tenant to generate token-url automatically    #tenant: azure-tenant-uuid    # OIDC client secret - either this or certificate is required    secret: abcd    # List of OIDC scopes to request    scopes:      - api://abc345/.default  # Data set to attach uploaded files to. Either use CDF IDs (integers) or  # user-given external ids (strings)  data-set:    id: 28834    #external-id: File_Extractor# (Optional) Extractor performance tuning.extractor:  schedule:    type: interval    expression: 5m# Information about files

Hi,I am setting up MFA on my new device and going through this https://cognite.zendesk.com/hc/en-us/articles/360019771298-How-to-Sign-up-for-Cognite-Hub The images are missed in this, can you please help me to send the details. Thanks.  

OEM’s are focusing on performance based contracts but are lacking information about their installed assets . It’s important to create a data foundation to map all assets . How does Cognite provide solutions ,is it by integrating with SAP /Maximo ?

Hello,As per your documentation for the SkipState behaviour:'In case of failures, retries will be performed until exhausted, after which the task is marked as FAILED and the Workflow is marked the same. In the event of a timeout, no retries are undertaken; the task is marked as TIMED_OUT and the Workflow is marked as FAILED.'But this is not what happens:  

CogniteAPIError when listing eventsI have observed an error raised when listing events that I have never seen before. Below is the full traceback 2024-10-28 03:35 : Traceback (most recent call last):2024-10-28 03:35 : File "/app/function/***/***/***.py", line 95, in get_events2024-10-28 03:35 : events = self.client.events.list(2024-10-28 03:35 : ^^^^^^^^^^^^^^^^^^^^^^^^2024-10-28 03:35 : File "/user_env/lib/python3.11/site-packages/cognite/client/_api/events.py", line 817, in list2024-10-28 03:35 : return self._list(2024-10-28 03:35 : ^^^^^^^^^^^2024-10-28 03:35 : File "/user_env/lib/python3.11/site-packages/cognite/client/_api_client.py", line 532, in _list2024-10-28 03:35 : for resource_list in self._list_generator(2024-10-28 03:35 : File "/user_env/lib/python3.11/site-packages/cognite/client/_api_client.py", line 463, in _list_generator2024-10-28 03:35 : res = self._post(2024-10-28 03:35 : ^^^^^^^^^^^2024-10-28 03:35 : File

I'm seeking help and guidance from customers who have tackled or enabled features to improve the LINES when connecting files and assets within Cognite Canvas.We seek to have our connected lines more accurately flow and follow P&IDs as they are added to the canvas.  Making the lines conform and adjust vs. extending across the visuals on the canvas.This means these lines should follow the “path” on the connected P&ID where they begin on the document vs. connecting randomly into the side of the document. 

When using Co-pilot to interrogate a document, it returns an answer and links to the page of the document from where Co-Pilot inferred the answer.  If you click any of those links, the document jumps to that page, but Co-Pilot disappears.  You cannot get back to the co-pilot pop-up to click the next link.  If you ask the question again, you get a different answer.  The original prompt, answer, and links need to remain visible and clickable until the user closes the window or asks another question. 

Are there any way of using the javascript SDK without storing access tokens in local storage (or similar)? We would really like to store access tokens in a location where the browser’s JS runtime don’t have access (such as an http only cookie). If not, are there any plans of supporting that?

We have an OPCUA Server (Server) that sends an array (Array) of length 500 every second. We need an OPCUA Client (Client) to stream this array to cloud.First, we use UaExpert an open-source software as Client. Here we can see a node for Array which has values of Array[0], Array[1], Array[2], Array[3], … Array[499] Next, we use Cognite OpcUaExtractor as Client to stream data to cloud. But we result in only Array[0] being sent in each instant (second). I have tried to adjust different parameters in config file with no success.

In a hypothetical scenario where we need to extract asset hierarchy directly from PDF-based P&ID documents, does Cognite provide a built-in library to support this? Or would we need to rely on external libraries like PyMuPDF or pdfplumber for data extraction—or even tools like Pytesseract in cases involving scanned images?Thank you for any guidance on best practices or integrations for this use case!

We are looking to add Events into our CDF solution via an existing OPC UA Extractor. The OPC UA Extractor is currently pulling data from KEPServer and pushing into time-series end-points.We want to add some of the binary & string references in the PLC (which are now exposed in KEPServer) as Events so that we can utilize the Charts over-lay function.Is there a worked method for updating the OPC UA Extractor file for this purpose? Does it require KEPServer Alarms & Events plugin to be used? We have used this link already: Configuration settings | Cognite Documentation

Hello everyone in the community,I have a question:I have a project created with CDF and the assets are already contextualized with the P&IDs. I received new revisions of plans.My question is, do I have to contextualize everything again and what has already been contextualized is lost? Is there a way to only upload the new plans in the new revisions and have it done automatically?

I am looking at 127 time series linked to one asset and I want to download the list of these time series as shown in the screenshot below, but this doesn’t appear to be straight forward.The download button circled in blue saves a JSON file linked only to the asset “11. QHP”. Is there a way to spare the user the effort of manually selecting and downloading each of the 127 timeseries and later reassemble them into one table like the one shown in the browser?  It is not possible to select and display more than 20 columns in the browser … due to performance issues. This is not critical at this time, but still dissatisfying. I want to download everything wholesale and pick what I need from the list locally. Is there a way around this restriction? Solving issue #1 would remedy also #2, as I’d be able to join the tables locally again. Thanks

Search option is not working in non-pdf document at CDF for Documentum.Steps to reproduce the issue: Login to CDF Click on Data Explorer tab in CDF menu bar. Click on Files tab in right side of the panel. Set Data set as ‘src:006:documentum:b60:ds’ under Common filters in left side of the screen. Enter the file name ‘ELECTRICAL SAFETY RULES.docx’ in search bar. Click to open the file. Expected results: User should be able to see the preview of file and search the any word inside the document.Actual results: Search option is not working in preview of file and displaying message ‘this document is not searchable.’

Hello Cognite Community,I am working on a workflow that involves receiving a daily email with a consistent title in Outlook. Each email contains an attachment that I need to extract and process. I'm looking for a way to automate this extraction process using Jupyter Notebook.Has anyone implemented a similar solution or could guide me on how to approach this? Specifically, I’d like to:Identify emails by their subject line Download the attachment(s) automatically Work with these files within a Jupyter Notebook environment for further analysisAny suggestions, code samples, or libraries that could help with this integration would be greatly appreciated!Thank you in advance for your insights

Happy Friday!We wanted to end the week with the news that both File Download and search & filtering in Tree view is now available in Search, see below product update post: Worth noting; the search & filtering ability in Tree view is available for projects with asset-centric data modeling. For projects where the data is modeled as an industrial knowledge graph, the same functionality will be made available as part of the official Q4 release December 3. Best, Sofie Berge, Product Manager

 Hello,  We are unable to select version of workflow from the UI. for instance here we are at version 0.0.39 and in the UI the workflow is at version 0.0.1.I think this might be an UI issue. Thanks