The 2026.R3 release brings AI closer to the work itself. The new TrendMiner Agent sits inside the application, understands what is on your screen, and helps you build an investigation as it develops. Three new MCP tools give external AI assistants deeper reach into your context data and monitor triggers. For teams who have been experimenting with AI alongside their process data, this release turns pilots to productized solutions. Beyond AI, 2026.R3 sharpens tools you already use: logarithmic scale support across TrendHub, a redesigned Query Designer for external data sources, and a context foundation rebuilt to stay fast as your history grows.
The TrendMiner Agent: AI where you already work
Connecting TrendMiner to the AI assistant your organization uses was the first step, delivered through the MCP Server. This release adds the second: an agent native to TrendMiner itself.
What's new? The TrendMiner Agent is embedded directly in the application, working alongside you in the browser. It understands the view you are on and helps you build an investigation as you go.
On the analytics side, the Agent can navigate asset structures, search tags and index information, run value-based searches to find events matching specific criteria, fetch tag data, calculate aggregations, run deeper time-series analyses, and search context items and context types.
What makes it different from a chat window on the side is the interactive layer. The Agent can open views and navigate within TrendMiner, add relevant tags to your TrendHub view, change the time range, and search or filter a context view. It acts on the application in front of you rather than describing what you should click next.
It also works as your in-app TrendMiner expert. Practical “how do I” questions, such as how to build a formula or how a feature behaves, get answered in context instead of sending you off to browse documentation.
Bring your own model. The Agent connects to your own LLM, using providers your organization already trusts. Azure AI Foundry and Anthropic are supported at launch, with AWS Bedrock and OpenAI planned. There is no new vendor to onboard, and your data stays within your company landscape.
Why it matters: This is the first release of the Agent and the starting point for agentic capabilities in TrendMiner. It shortens the distance between a question and an answer for engineers who know their process well but do not spend their day inside analytics software.
The TrendMiner Agent is available today on request. Reach out to your TrendMiner contact to get started.
MCP Server: three new tools, more context to work with
What's new? The TrendMiner MCP Server connects your process data and analytics to the AI assistant your team already uses, including Microsoft Copilot, OpenAI ChatGPT, or Anthropic Claude. This release expands the toolset with three additions:
- search_context_types. Your AI agent can discover which context types exist in your environment, such as batches, campaigns, or shutdowns, before querying them. It asks better follow-up questions instead of guessing at naming conventions.
- get_context_item. Pull the details of specific context items into the conversation. An operator can ask when the last maintenance on Pump 12 took place and what was done, and get start and end time, associated tags, and notes without leaving the chat window.
- list_monitor_triggers. List which monitors have triggered on a tag and when. A process engineer asking which alarms fired on Reactor 3 last week gets a direct answer rather than working through the monitor overview.
Why it matters: The first MCP release covered discovery and time-series analysis. This one adds the operational context that makes those answers trustworthy. A tag value on its own rarely explains much. The same value placed against a shutdown, a campaign, or a maintenance event usually does.
The MCP Server is available today on request. Reach out to your TrendMiner contact to get started.
Logarithmic scale support in TrendHub
What's new? Some process values span several orders of magnitude, and on a linear axis the small values flatten into a line along the bottom of the chart. You can now set the scale type per tag to linear or logarithmic, alongside the existing automatic and manual range settings.
Your choice is remembered for your user and stored in the view, so dashboard tiles built on a saved TrendHub view show it too. The scale type applies across TrendHub: trend and stacked trend plots, the scatter plot, and histogram bins. It can also be applied to a group of tags at once.
Zero and negative values are still plotted. TrendMiner uses a symmetric logarithmic scale, so no data points drop off the chart.
Why it matters: Reaction kinetics, conductivity, particle counts, trace concentrations. Anywhere the interesting behavior happens at the small end of a wide range, a linear axis hides it.
Query Designer for Generic ODBC and JDBC data sources
What's new? Configuring a Generic JDBC or ODBC data source previously meant committing to every query upfront inside a single wizard, with no way to check results before saving. The new Query Designer separates connection setup from query building, so ConfigHub administrators can configure a data source and then build, test, and refine each query on its own.
- Save a data source in draft while you work through its queries, and activate it only once everything checks out.
- Fill in template variables and run a live preview of each query before saving, with validation feedback the moment something is off.
- Switch between queries without losing preview results or variable values.
Why it matters: Fewer round trips between configuration and verification, and less risk of a half-correct data source going live.
A context foundation built to grow with your history
What's new? Context data storage has been restructured into yearly partitions, so queries on recent data stay as fast years from now as they are today. Nothing changes in how you use TrendMiner day to day.
Built on that same work, new integrator endpoints let external systems ask a more precise question: what has changed since I last checked. Integrations with Power BI, Databricks, and other systems in your data ecosystem can sync incrementally by creation or last-modified date, which fits how those systems already think about polling for updates.
Why it matters: Plants accumulate context. Batches, shutdowns, quality events, shift notes. The value of that history only goes up, and performance should not go down as it accumulates.
New license format
In preparation for upcoming versions that will provide license assignment and enforcement, a new license format needs to be installed before the release of TrendMiner 2026.R4.0. It includes the contractual boundaries covering user types and system quotas.
You can install it at any time after upgrading to 2026.R3.0, and it is a required step before upgrading further. After uploading, entitlements can be verified in EdgeManager. New licenses will be distributed before the next release. If you would like yours early, contact your TrendMiner representative.
For detailed information on the release, please refer to the release notes.
