What's New at spektr

Features

Version history for SummaryAI

You can now view previous versions of a customer's AI-generated summary. A date dropdown on the SummaryAI block lets you switch between past versions, with a "See latest version" button to jump back to the current summary. This gives compliance officers and auditors visibility into how the AI's view of a customer evolved over time.

What this changes for you: No change to your current workflow. The summary displays as before, and version history is an optional dropdown you can use when you need it.

Voting rights on DnB ownership insights

You can now see voting rights percentages for customers where this data is available from DnB ownership insights, shown in a new field alongside the existing ownership percentage. You can also filter customers by voting rights percentage, choosing whether to combine it with the ownership percentage filter using AND or OR logic.

What this changes for you: No change to existing workflows. This adds a new field and filter option for teams using DnB ownership insights.

Research AI source trust labels

Each source in Research AI's results now carries a trust label of either Trusted or Needs verification, generated automatically by an LLM-based evaluation when the research runs. Sources are shown with a green (Trusted) or amber (Needs verification) badge so you can gauge source credibility at a glance without leaving the page.

What this changes for you: No change to existing workflows. This adds a visual cue to help assess source reliability faster during research review.

Batch evidence upload

You'll be able to upload multiple evidence documents to a case in a single batch instead of one at a time. The upload dialog will support selecting or dropping multiple files at once, show per-file upload progress, and let you retry just the files that failed. This is rolling out gradually behind a feature flag.

What this changes for you: No change today. Once enabled for your workspace, uploading multiple documents to a case will be faster since you won't need to upload files one at a time.

Improvements

Aggregation node filtering

The Aggregation node now supports filtering, so you can aggregate over only the related entities you want, such as excluding archived ones, instead of always aggregating over all of them.

Discrete CPR address fields

The CPR (Danish civil registry) integration now returns street name, house number, zip code, and city as separate fields, in addition to the existing combined address field. The CPR event details in the timeline show these fields individually as well. This makes it possible to monitor address changes more precisely, such as catching a real change in house number rather than relying on a fuzzy match across the whole address string.

License AI empty source guard and license count output

License AI no longer returns spurious results when the source field it checks against is empty; it now returns no findings instead. It also always outputs a license count number, which can be connected to downstream risk or calculation nodes: this counts all active licenses found in Discover mode, or only the confirmed licenses in Confirm mode.

License AI source limit

License AI now limits how many outside sources it can reference in a single run, capped at five, so it can no longer keep searching indefinitely when it doesn't find what it needs.

This release also includes general performance improvements and bug fixes.