Skip to content

Release 7.2

October 2026

Whats New In Our Products ?

Kelvin UI

What's New

Data Explorer 2.0 (KPD-155, KPD-207, KPD-275, KPD-297)

  • Activity Lanes. Events, Insights, Recommendations and Actions (control changes and custom actions) each have a lane on the chart and a searchable, filterable card list in the side panel. Hovering or clicking a card highlights the matching point on the chart, and vice versa.
  • Events in Data Explorer. Pending unsupervised and similarity events are shown on the chart for the first time, and can be ignored or turned into insights.
  • Details beside the chart. Clicking an activity opens its details next to the chart, with a time marker or band showing where it sits. Recommendations can be accepted or rejected from there.
  • Saved views. A DE View picker saves the chart setup (data streams, aggregations, Y-axis, time range, lanes and filters) as named views, shared with all users or kept private. Kelvin View is the default.
  • Y-axis range per data stream. Set a minimum and maximum per data stream, and see each stream's unit. Also available on Data Explorer chart widgets.
  • Add Insight from the chart. Select the range with handles on the chart and fill in the form in the side panel. New insight types can be created from the same form.
  • Assets List scope. The asset list can follow a saved view from the Assets table.
  • Compare up to four assets in multi-asset mode, with every card showing which asset it belongs to.
  • Activity on the inline chart. The chart in an expanded Assets row shows activity lanes and a view picker; clicking an activity shows a summary with View More into Data Explorer.
  • Links to an activity. A link can open Data Explorer directly on a recommendation, action or insight.
  • Legend. Double-click a series to show only that one.
  • Snooze from Data Explorer (see Snooze below).

Snooze recommendations and insights (KPD-258)

  • Stop a Recommendation Type or Insight Type from surfacing on an asset for 3, 7, 14, 30 or 60 days, or until a chosen date.
  • Snooze from the Assets table, the Asset Detail header, the recommendation preview and modal, the insight preview chart and Data Explorer.
  • The Snooze Configuration modal edits every type for an asset at once, one by one or in bulk, and shows how each change extends or shortens the snooze.
  • A banner marks a snoozed type, with Edit Snooze to change or remove it.
  • Recommendation and insight columns on the Assets table gain a Status filter that includes Snoozed.

Comments (KPD-276)

  • Comments on setpoint changes and parameter changes.
  • A Comments panel in the Action Details modal, which now also shows parameter changes.
  • A Comment column on the Assets table showing the latest comment per entity type.
  • Mentions and live updates in recommendation comments.

Insight Types management (KPD-303)

  • An Insight Types tab on Engines → Insights to review, create, edit and delete insight types, including their color.

Assets

  • Decimal places can be set per data stream column (KPD-334).
  • A Tasks widget on the asset Control Panel lists the tasks running on the asset.
  • Multi-select dropdowns support Shift+click to select a range of options.
  • Connection IO mapping can be filtered and sorted by asset and data stream.

Administration

  • Role cards in Roles & Permissions show the role description (KPD-322).

What's Deleted

  • Auto Scale Y-Axis toggle in the Data Explorer header. Charts autoscale by default; set a minimum of 0 on a data stream to get the old "off" behaviour.
  • Same Asset Type scope option in the Data Explorer asset list. The scopes are now All Assets, Assets List, Task and Application.
  • Data Source / Activity buttons in the Data Explorer header. Use the tabs at the top of the side panel.
  • Default time range setting in Instance Configs → Data Explorer. Set the time range on a shared view instead.
  • Data Explorer configuration saved per asset type. It is no longer applied, and is not converted into views. Charts open on Kelvin View or the selected view.
  • Opening the Insights page from Data Explorer to create or edit an insight. This now happens inside Data Explorer. Creating an insight from an event on Engines → Events is unchanged.
  • Attachment icon in the recommendation comment box (attachments were never supported).

What's Deprecated

  • The "Data Tag" name. It is being replaced by "Insight" everywhere in the product.
  • Data Explorer settings stored per user or per asset type (core.ui.data-explorer.settings.asset-type, and the lane visibility, Fill Data, refresh interval, auto-scale and time range settings). These now live in views. Existing records are left in place but no longer read.

What Changed

  • Data Explorer layout. The side panel is on the left and always open. Assets and data streams are in one Data Source tab, filters open inline, and the data streams section lists every data stream, with the selected asset's first.
  • Recommendation and action details in Data Explorer open next to the chart instead of in a modal.
  • The first asset is selected automatically when Data Explorer opens without one, or when the scope changes.
  • Edit Aggregations is now Data Streams Configurations, and its button reads Apply.
  • Engines → Insights now has two tabs; the list moved to /engines/insights/matches (old links still work).
  • One leave prompt everywhere. Every form, modal and side panel with unsaved changes asks "Are you sure you want to leave?", with Cancel / Confirm.
  • One add-comment modal for accepting or rejecting a recommendation and confirming parameter changes, with a visible character count. Shift+Enter adds a new line.
  • Manual Controls History is a timeline grouped by day, showing who made each change and any comments.
  • Applications column on the Assets table shows each connector instance separately, with a +n pill for the rest.
  • Bulk delete sends every selected item and reports which were deleted and which were skipped, with See Details. Disabled delete and edit buttons say why.
  • Connections deploy the data quality metrics their connector declares.
  • Task workloads open under their task. Opening a workload under the wrong kind of page shows Page Not Found.
  • Asset Map zooms in further.
  • Docker applications can be deployed before any assets exist.

Fixes

  • Asset types in use can no longer be deleted, and the tooltip lists what uses them. A deleted asset type no longer stays stuck in the Assets filter.
  • Large Connection IO mappings (10,000+ rows) no longer show an error after refreshing, and the tab is faster.
  • Decimal values can be typed in numeric IO mapping columns; editing an OPC UA IO mapping no longer fails.
  • Cancelling a connection from the data quality flow works.
  • Data Explorer: lines with more than 5,000 points are drawn when the browser has no WebGL; opening a recommendation no longer changes the chart's aggregation; the data streams list loads faster.
  • Last Insight previews no longer turn into an empty state after polling or before the chart loads.
  • Tables no longer reload completely when returning to an inactive browser tab, and columns with lists no longer overlap.
  • Users no longer silently lose saved settings because of corrupt stored records; a version-mismatch error when saving a setting is fixed.
  • Fewer crashes from unexpected data: the Assets filters, a missing data stream on a control-change card, the control mode widget, the Add Assets wizard and the Applications page.
  • Each page load now makes one permission request instead of about 20.
  • Logout no longer errors, and logging in from a Grafana tab no longer loops to logout. Refresh tokens are no longer sent to error reporting.
  • "Something went wrong" is replaced by the actual reason after a resource is deleted.
  • Impact: asset search in the metric preview filters correctly; economic parameter names are validated while typing; Save is no longer enabled after removing items.
  • Tasks: downloaded CSV includes parameter changes; validation errors show in full on hover; navigating from a task workload to Data Explorer keeps the right context.
  • Data type is no longer missing when creating a data stream.
  • Workload deploy no longer shows an error while checking the name.
  • Layout fixes at 1280 px: Accept/Reject buttons, search field overlap, missing-conditions column width and counter.

Kelvin API

What's New

  • Snooze insights and recommendations: you can silence a data tag or a recommendation type on one asset until a date you choose, without turning off the app that raises it. Apps respect your snoozes.
  • Comments in one place: you comment the same way on recommendations, control changes and insights, and you can show the latest comments next to each asset.
  • Data Explorer 2.0: events, insights, recommendations and actions each load in their own lane. Lanes load faster, and search covers every page of results.
  • Custom action results: the system that carries out a custom action can report back, for example with a work order number, and you see that result in Kelvin.
  • Filter control changes by app: one filter shows every change an app made, across all its versions and deployments.
  • Smarter app deployment: apps can tell Kelvin when they are ready to take work, read secrets as protected files, and send their own metrics to monitoring with no extra setup.
  • Clearer parameter history: Kelvin groups parameters you change together into one entry, so you see who changed what and when.
  • Asset type usage: each asset type lists the assets that use it, so you can see why you can't delete it yet.
  • Cleaner data: Kelvin rejects recommendations and control changes for assets that don't exist. Items left over from deleted assets show up in lists, so you can clean them up.
  • Cluster inventory: you can see what runs on each edge cluster from the cloud, without logging in to the site.
  • Faster edge upgrades: an upgrade only touches the edge services that changed.
  • Installs behind strict firewalls: Kelvin downloads everything from its own registry, so a site only needs to allow Kelvin domains.
  • Kubernetes upgrades in place: you can upgrade a single-server cluster to a newer supported Kubernetes (k3s) version without reinstalling. Kelvin blocks this on multi-server clusters to keep them safe.
  • Request tracing: support can follow a slow or failing request across services and find the cause sooner.
  • More tools for AI assistants: AI agents connected to Kelvin can work with impact, snoozes, properties, data tags and permissions. They also learn the size limits up front and get clear advice when a request is too large.
  • Permissions guide: the documentation shows how to limit a role to specific resources, such as a single well.

What's Updated

  • Breaking changes for integrations
    • Integrations that create or list threads need an update: thread creation takes a new request format, and thread lists come back in pages.
    • Control changes no longer accept structured (object) values, and only known sources can create them.
  • Fixes
    • Control change status updates no longer fail or lose their state.
    • Apps no longer redeploy on their own after frequent updates.
    • Updating a data stream keeps its type.
    • Impact calculations keep running when they meet unusually large values.
    • Asset insights no longer return errors when you page through results.
    • Snoozes work for assets whose names contain special characters.
    • Date and time filters work again in lists that use them.
    • Kelvin checks custom action IDs on update and stops repeating cluster syncs.
    • Recommendations no longer clash when several change state at the same time.
    • The data flow service survives a dropped connection.
    • Downloading an empty result no longer gives you a broken file.
    • Saving user settings retries on its own when two saves collide.
    • Cloud and edge reconnect cleanly when the link drops.
    • The latest-values query handles a very large number of data streams.
    • Login sessions follow the configured maximum length.
    • k3s installs and multi-server cluster joins succeed more often.
  • Performance and security
    • Asset insights and recommendations load faster.
    • Asset status updates write less to the database.
    • Kelvin runs on Go 1.27 and current versions of key libraries, which include security fixes.

What's Deprecated

  • Timeseries "type" field: use "data type" instead. Kelvin accepts both for now and uses "data type" when you send both.

What's Deleted

  • Old thread actions: following a thread, marking it as seen and replying are gone. The new comments replace them.
  • InfluxDB: Kelvin now stores time-series data in ClickHouse only.

Kelvin SDK

What's New

  • Control Change Manager rewritten in Go: Including Control Changes Sync Logic that was previously part of Edge-Sync.
  • Modbus connector rewritten in Go: It covers RTU serial and TCP, with the same behavior and a smaller footprint.
  • Leader election for HA app replicas (kelvin.ha): Apps can run multiple replicas with exactly one active, with no configuration required.
  • Liveness from code (kelvin.health): Apps can report their own liveness.
  • Readiness probes: readiness_probe is supported alongside liveness.
  • Secret file volumes: Secrets can be mounted as read-only files.
  • App metrics endpoint: Apps can declare a Prometheus endpoint for the platform to scrape.
  • Data Quality modules are now supported by importers and exporters.
  • Custom Action Responses: Apps receive the result of the custom actions they dispatch.
  • Insight and Recommendation snooze: Adds a snoozed recommendation state and a new Insight object.
  • OPCUA Error Data Quality metrics: Status codes such as BadSensorRead are published per asset and datastream.
  • Microsoft Teams exporter.

What's Deleted

  • Legacy Modbus connector: It is replaced by the Go version.
  • api-client thread replies methods: They are replaced by comments.

What's Deprecated

  • DataTag: It is replaced by the Insight object.

What's Changed

  • The api-client is updated to the latest platform spec.
  • App descriptions are limited to 256 characters.

Fixes

  • The gateway rejects null payloads instead of crashing, and the SDK blocks None payloads.
  • The gateway reconnects after a websocket closes.
  • OPC UA retries failed monitored items with backoff.
  • Each OPC UA metric that share a node ID receives data.
  • kelvin workload logs no longer crashes on log lines with bracketed paths.
  • kelvin workload logs -o overwrites the output file instead of appending to it.
  • kelvin app agent sync can no longer write files outside the app directory.
  • A failed kelvin app mlflow create or import overwrite no longer deletes existing files.
  • kelvin app create quotes descriptions, so the generated app.yaml stays valid.
  • kelvin auth login builds a valid OAuth URL when the authorization endpoint already has query parameters.
  • kelvin app images unpack extracts filenames that contain ...
  • Docker buildx app uploads work again when image layers are shared.
  • The api-client no longer drops aliased request fields such as schema and from.
  • The api-client's chunked DataFrame conversion stops at the end of the input instead of repeating data.
  • The api-client's retry transport handles non-standard HTTP status codes such as 599.
  • Publisher parameter overrides no longer leak between instances.
  • The publisher keeps False and 0 parameter values instead of replacing them with defaults.
  • The publisher rejects invalid messages instead of reporting them as published.
  • Window buffers keep timestamps in order.
  • Time quantization no longer loses a microsecond.
  • A malformed app section or an unsupported YAML tag now raises ConfigError.

Whats New in our Documentation ?

The Data Explorer has been rebuilt for Data Explorer 2.0, and its documentation is rewritten from the ground up. The Overview and Monitor an Asset's Data pages describe the new layout, and new pages cover Custom Views, Events, Insights and Actions. The Recommendations and Download Data pages are updated to match, and the old Control Changes and Tags pages are replaced.

Recommendation types and Insight types can now be snoozed for an Asset. The Recommendation, Insight and Managing Assets pages explain how to snooze from the Assets table, the recommendation popup and the Data Explorer, and how to edit or remove a snooze. Insights now has two tabs, with new Insight Matches and Insight Types pages.

Developer Tools gains new pages on exposing Application metrics, health and readiness, leader election, mounting a secret as a file, and downloading comments. The Custom Actions pages explain how an Application sends and receives a response, and the Insights pages replace Data Tag. Platform Administration adds Cluster Inventory, a Kelvin Provision CLI reference and an Interpreting Built-in Metrics page, and uses the Data Quality Check and Metric terms throughout.

71 short demo videos now sit beside the written steps across Operations Tools and Platform Administration, covering the most common tasks step by step.

What's New

What's Updated

What's Deleted

  • Operations Tools ⟶ Data Explorer ⟶ How To ⟶ View Control Changes page
  • Operations Tools ⟶ Data Explorer ⟶ How To ⟶ Create/View Tags page

Others

  • 71 new demo videos added across 55 documentation pages, covering assets, asset types, data streams, connections, guardrails, clusters, nodes, Data Quality, AI Events, Insights, Tasks and the Data Explorer
  • Screenshots refreshed to the latest look and feel across Operations Tools and Platform Administration
  • Asset Management, Data Streams and Connections now reflect the Connectors menu on the left sidebar
  • Update renamed to Edit for Guardrails and Data Streams
  • Data Quality Checks, Applications and Metrics terminology aligned across the documentation
  • All Data Explorer references across the documentation updated to Data Explorer 2.0
  • Beta tags removed from the AI Tools and Tasks pages

Kelvin Academy

The Kelvin Academy courses and Kelvin Shorts are updated for Data Explorer 2.0, snooze, Data Quality Checks and the new Connectors menu. They also embed the new demo videos beside the written steps.

What's New

  • None

What's Updated

  • Kelvin Academy ⟶ Index page
  • Kelvin Academy ⟶ Developers ⟶ Intermediate ⟶ 21X Developer Courses Introduction page
  • Kelvin Academy ⟶ Developers ⟶ Intermediate ⟶ 210 Quick Start page
  • Kelvin Academy ⟶ Developers ⟶ Intermediate ⟶ 211 Applications page
  • Kelvin Academy ⟶ Developers ⟶ Intermediate ⟶ 212 Connectors page
  • Kelvin Academy ⟶ Developers ⟶ Intermediate ⟶ 223 Recommendations page
  • Kelvin Academy ⟶ Developers ⟶ Intermediate ⟶ 224 Data Quality page
  • Kelvin Academy ⟶ Developers ⟶ Advanced ⟶ Roles ⟶ Data Scientists ⟶ 310 Databricks MLFlow page
  • Kelvin Academy ⟶ Developers ⟶ Advanced ⟶ Roles ⟶ Data Scientists ⟶ 311 Databricks Volume Uploader page
  • Kelvin Academy ⟶ Operations ⟶ Intermediate ⟶ 240 Exploring the Kelvin UI page
  • Kelvin Academy ⟶ Operations ⟶ Intermediate ⟶ 241 Adding Assets to Applications page
  • Kelvin Academy ⟶ Operations ⟶ Intermediate ⟶ 242 Asset Table View page
  • Kelvin Academy ⟶ Operations ⟶ Intermediate ⟶ 243 Review and Approve Changes page
  • Kelvin Academy ⟶ Platform Admin ⟶ Intermediate ⟶ 27X Platform Admin Courses Introduction page
  • Kelvin Academy ⟶ Platform Admin ⟶ Intermediate ⟶ 270 Exploring the Kelvin UI page
  • Kelvin Academy ⟶ Platform Admin ⟶ Intermediate ⟶ 271 Setting Up Platform Infrastructure page
  • Kelvin Academy ⟶ Platform Admin ⟶ Intermediate ⟶ 272 Setting Up Assets and Data Streams Quickly page
  • Kelvin Academy ⟶ Platform Admin ⟶ Intermediate ⟶ 273 Asset Data Connectivity page
  • Kelvin Academy ⟶ Platform Admin ⟶ Intermediate ⟶ 274 Asset Data Assurance page
  • Kelvin Academy ⟶ Kelvin Shorts ⟶ Operations Tools Shorts page
  • Kelvin Academy ⟶ Kelvin Shorts ⟶ Platform Admin Shorts page

What's Deleted

  • None

Others

  • 71 demo videos embedded across 11 Kelvin Academy pages and in Kelvin Shorts
  • Data Quality Checks, Applications and Metrics terminology aligned with the documentation