Product User Community

Hi Cognite Community,We are looking at ingesting approximately 1 million engineering and operational documents into CDF, including equipment manuals, datasheets, inspection reports, SIL/HAZOP studies, operating procedures, technical studies, maintenance reports, and other asset-related documents.Our key requirement is for an Atlas AI agent to answer questions across this entire document repository without the user having to identify or pinpoint a specific document first.For example, a user should be able to ask:“What are the recurring compressor seal issues reported across our engineering and operational documents?”The user may not know which documents contain the relevant information. We therefore expect Atlas AI to discover the appropriate documents/passages, retrieve the relevant content, and generate a grounded response with references to the source documents.According to Cognite's documentation, CDF Semantic Search does create vector embeddings for PDF documents as part of the bui

Whether CDF natively supports semantic/ontology-based relationship modelling (named, typed relationships like contains, hasTimeSeries, performedOn with direction/cardinality/metadata), and whether Atlas AI and other consumers can traverse that semantic meaning rather than just node-to-node links.

In order to delete nodes in CDF in a given View because of deletes in the Fabric source table, we need to extract necessary fields for all rows in Fabric to construct the space and ExternalId, and then compare this with a full extraction of the space and externalId in the View.In one example with ~400k rows/nodes, it takes 1-2 minutes to get the data from Fabric. However, the cdf_nodes is several orders of magnutide slower.Only extracting space and externalId for 400k nodes in the View takes ~25 minutes Adding a Where clause to check if the externalId has a given prefix as a safeguard makes the query time out.Using cdf_data_models or other functions does not seem to help either. Is there a reason that listing ids for nodes is so slow? And are there other ways to go about this use case?

Hi Cognite team,>> On project `essc-sandbox-44` (cluster `westeurope-1`), we've configured a published AI Agent's data tools with **"Follow access scope defined manually"** set to a single instance space (`beveragemanufacturer`), but the tools scan the entire project's cross-space instance population instead of the scoped space. We verified via the SDK that the agent's "Found data of type" counts match the global totals exactly (e.g., 31,120 Process Alarms and 39,260 Downtime Events project-wide) while the scoped space holds only ~334 instances total — producing 6+ minute responses. It reproduces across three different tool types and is unaffected by reducing the data model, removing tools, or adding explicit prompt-level scoping.>> The attached document has full environment details, the verified evidence, reproduction steps, and our specific questions. Our main question: is this scope setting meant to filter DMS retrieval by space, and if so, is the cross-space scan a know

Hi Cognite Community,We have a requirement to extend the CDF 3D experience beyond displaying attributes from the Core Data Model, so that users can visualize related transactional and operational information as configurable layers directly on the 3D model.Our industrial knowledge graph connects equipment and functional locations to multiple types of information, for example:SAP Notifications and Work Orders Maintenance Plans Inspection findings Reliability events / Bad Actors RCA and TFN records MOC records Alarms and operational events Engineering and operational documents Time-series conditions and calculated KPIsThe Cognite Core Data Model provides standardized industrial concepts and can be extended through additional models and relationships. [docs.cognite.com], [docs.cognite.com]However, we would like the 3D visualization to make greater use of these contextual relationships, rather than limiting the user experience primarily to the attributes directly associated with the core 3D

Hi Cognite Community,We have a requirement for CDF Charts to dynamically use configured tag limit attributes as chart thresholds, rather than requiring users to manually enter static threshold values on an ad-hoc basis.For industrial tags/time series, we maintain multiple engineering and operational limits, for example:High / High-High limits Low / Low-Low limits Operating upper and lower limits Design limits Alarm limits Safe operating limits Other tag-specific engineering thresholdscurrently supports configuring thresholds for time-series analysis and monitoring. However, we would like these thresholds to be dynamically sourced from the contextualized attributes associated with the tag/time series, rather than manually entering a value each time a chart is created.Expected capabilityFor example, if a pressure tag has contextualized attributes such as:High: 80 bar High-High: 90 bar Low: 30 bar Low-Low: 20 bar Design Pressure: 100 barWhen the user adds that time series to a CDF Chart,

Hi Cognite Team, We have been using DM subscriptions for a while now and we know that it is very common for subscriptions to stop responding and we are forced to create new ones or start from a fresh cursor multiple times.I have questions on how will this be monitored and responsibility for fixing it in case it doesn't work or stops responding and how do we find it. Can we have some alerts or emails sent to us incase if anything fails similar to for Extractors when it fails? cc: ​@Neerajkumar Bhatewara 

I’m unable to use the Microsoft Authenticator app on my new phone, as my previous phone has been sold.I need to access the public data instance for the Cognite Flows workshop, but authentication is required to log in. Since I’m unable to receive or approve the authentication request on my phone, I’m currently blocked from accessing the instance.Could you please help me resolve this authentication issue so I can access the public data instance and proceed with the workshop?

In Flows (Search and Charts), a Data Modeling time series shows up under the Default categories (Classic data model time series) by default. When selected in Search it appears with incomplete metadata (including an empty External ID) and cannot be added to a chart (“You can only add time series with external IDs to a chart”), even though the underlying time series does have an external ID.The same time series can also be found under Other categories, where it works as expected: the External ID is shown and the time series can be plotted.The confusing part is that users are presented with the non-working version by default, and there is no obvious indication that they should instead select the time series from another category. Since both entries appear to represent the same time series, I would consider this a UI bug or an inconsistency between how Classic and SDM time series are surfaced in Search & Charts.  

Hello! I have got a new phone and am not able set up the authenticator app again.Could you please help with the set up.Thanks,Rahul

Team,We are having an issue where a customer can login from certain geo locations but not others.Where access is denied, we get a 403 response. The customer’s IT team is involved, and they can’t seem to solve this issue.Has anyone experienced this issue?

When i login to Congite-learn organization, i have access only project cdf-fundamentals when i switch to another project. I don’t access. Please advise and help how to get it.

As we move towards having a data model and location filter we consider as the default for our company we no longer see a use for the system default in most our CDF projects. Is there a way to override the system default so that all new users have one of our custom location filters rather than system default.We are moving away from CDF classic, and we only use the core data model by extension, so the system default causes more confusion at this point than anything else.Rather than “System default”, a “Default” set by ourselves is preferred. Markus PettersenAker BP - Data Platform Architect

Minor UI issue.In Charts, Time series preview often appears messy with overlapping text.  Two examples:   

What we need, and why it is per-asset.We ingest from OPC-UA. One extraction pipeline serves a production line carrying several machines. The PLCs publish on change, so a tag going quiet is completely normal — and from the data alone it is indistinguishable from the source having stopped.Our calculations forward-fill. So before forward-filling a value we have to know up to what point it is safe to do so: the last moment we know that machine was actually reporting. That is a property of the machine, not of any individual time series — every tag on a machine shares the same liveness — so what we need is one current value per asset, kept up to date continuously and read on every calculation.How we did it with classic resources. We wrote a Last Synchronized key into the classic asset's metadata bag. One small write per asset per cycle, and it cost nothing against data-modelling limits.Why this is awkward in the core data model. CogniteAsset has no free-form property — none of its 23 propert

Hi team !Is it possible to configure a backup (failover) server for the PI Extractor, so data collection continues if the primary server goes down?Thanks!

We hit an undocumented behavior of the OPC UA extractor (version 2.38.0, Docker) that we'd like to confirm.Setup: cognite.metadata-targets.clean with assets: true, timeseries: true, relationships: true, plus extraction.update fully enabled. This had been running fine for months — classic asset hierarchy plus classic time series.What we did: we added the GA Core Data Model space option to the same clean block: metadata-targets: clean: assets: true timeseries: true relationships: true space: "<our-instance-space>"What we observed:As documented, time series switched to CogniteExtractorTimeSeries instances in the space, with datapoints available via instanceId. Not documented: classic asset creation stopped entirely, even though clean.assets: true was still set. We verified this by deleting the extractor-managed asset hierarchy and restarting the extractor: after a full startup browse, no assets were recreated — no error, no failure run on the extraction pipeline, nothin

Who we areWe run a web application for industrial bakery lines that uses CDF as its back end. It's used by both plant operators and management. We're currently migrating our data models to the Core Data Model, and while doing so we've extended it with views of our own — which is what we understood a data model to be for.While migrating we've hit a case where two things we need seem to be mutually exclusive, and we'd like to know the intended pattern.What we needTranslations for some units. Not all of them — a specific set where the label carries meaning for the reader. Our headline case is throughput. We measure, depending on the line and the moment, trays per hour, where a tray may hold several products, or baguettes per hour, or products per hour. #/h is correct dimensionally but doesn't express that variety, and our users need the right word in the right language. (A secondary, milder case: length:kilom has longName: "Kilometer", and there are no accented aliases in the catalogue —

To all the experts out there,When parsing a diagram, some annotations may be incorrectly assigned and overlap with the correct ones. For example, "Pump A-1" is correct, while "Pressure Gauge A-1" is incorrect.I want to reject the incorrect "Pressure Gauge A-1" annotation, but I cannot select it because it completely overlaps with the correct "Pump A-1" annotation. As a result, I cannot select and reject the annotation I want to remove. First of all, it is difficult to tell at a glance which annotations are overlapping, as the only indication is a slightly darker color. The annotations I pointed out below are overlapping.I would like it to work like Explorer, as it used to.

Hi, could you please help me to register MFA using Cognite to do bootcamp prep hands on session?

The latest version of the Grafana connector supports unit conversion when CogniteUnit is defined. However, I’m not able to add the target unit to the label. I can add the defined unit using {{unit}}. Using the “Query Inspector” I see that the target unit is returned in the data object via the unitExternalId, but I cannot display it using {{unitExternalId}}.   Resulting legend Query inspector resultsI notice that some of the properties can be shown, like type and isStep, but I assume it is because I get it directly from the timeseries instance?Any suggestions to how we can add the target unit (without having to manually write it to the label)?

Hello!We have encountered some issues with backfilling on the PI Extractor. This usually goes automatically, and works fine. However, we do now have a 2 day gap in data on Cognite, however all that data is present in PI.Are there any way to “force” a backfill, or fix this issue somehow?Thanks! 

Hello,I have a user who is interested in seeing a rolling total of a value within our Cognite instance. We have created a chart where we have a very simple setup of the data point we are interested in seeing over time, and a 30 day sliding window integration function on it, which is working fine and showing the value within the chart as needed. However, we are also interested in writing this data to another time series and then being able to display the current value of this time series on a canvas using the Live data point mode. I saved & scheduled the calculation to run daily within the Chart but the scheduled calculation and the time series which it created is getting no data, with no errors anywhere that I can see. Is this something that is possible to compute within Charts?

 We're migrating a production line from classic time series to CogniteTimeSeries in the data model, ingested by the OPC-UA extractor from a Siemens IIH server, and we'd like to understand how much of the metadata we hold at the source can travel with the signal rather than being reapplied afterwards. Three things, in increasing order of interest to us: Engineering units — the UnitId is discarded. Our server sends the full EUInformation: NamespaceUri = http://www.opcfoundation.org/UA/units/un/cefact, UnitId = 4937544, DisplayName = kW·h, Description = kilowatt hour. What lands in CDF is only the composed display string, in classic unit and now in sourceUnit. The typed reference is never populated: across 2,772 OPC-UA-sourced classic series in our production project, unitExternalId is set on zero, and on the data-model side the unit direct relation to CogniteUnit is null on all 854 series of the migrated line. We fill it afterwards with our own script, but the UNECE common code in UnitId

We are migrating to the core data model. Time series we tested. We can ingest CogniteTimeSeries directly from the OPC-UA extractor. Assets are in progress.Sequences are our open question. We use them heavily for three things:Multi-dimensional rows tied to a timestamp or time range, where one time series per dimension is the wrong shape. Static reference data that many Functions read. Schemas where one data modelling instance per row would explode our instance count — short heating episodes, small production batches, and similar.What makes them attractive is the combination: readable in the UI, integer indexing, cheap, fast, and metadata slots both per column and per sequence.Looking at the core data model reference, I see no sequence concept — CogniteTimeSeries, CogniteActivity, CogniteFile and so on are all there, but nothing for tabular data. We have noted CogniteTimeSeries with type: state and CogniteStateSet, which covers state classification, but not the multi-column case.So:Is a