Release 2026.06: Bringing Data Observability Into Your Code

Data quality work has quietly moved out of the UI. Pipelines are defined in code, deployments are containerized, and environments are promoted through CI/CD. With Release 2026.06, digna meets teams where that work already happens.

This release introduces the digna Python SDK, official Docker support, a new QueryMode setting for SQL execution, a configurable prediction model for anomaly detection, a redesigned dashboard, and extended import and export for validation rules.

Watch the release overview: digna Release 2026.06 on YouTube

digna Python SDK: automate everything from code

The headline feature of 2026.06 is the new Python SDK. It is a single install away:

From there, most of what you would normally configure in the digna interface can be expressed in Python:

• Create and configure data projects programmatically

• Trigger inspections and monitoring executions

• Manage datasets, validation rules and configurations in code

• Profile tables and extract metadata insights

• Export profiling and data quality results to external systems and repositories

• Integrate digna with notebooks, orchestration tools and CI/CD pipelines

For data engineers, this turns data quality setup into infrastructure-as-code. Project definitions live in version control, review happens in pull requests, and a new environment can be provisioned from the same script that built the last one.

For data scientists, the SDK opens a different door. digna's anomaly detection and profiling results become inputs to your own analysis. Pull profiling history into a notebook, join it with business data, or feed quality signals into a downstream machine learning workflow. Observability stops being a separate dashboard and becomes part of the analytical toolchain.

And because digna runs entirely inside your environment, none of this changes the data sovereignty model. The SDK talks to your installation, not to ours.

Docker support: deploy digna where it makes sense

Release 2026.06 adds official Docker image support. Installation, testing, scaling and environment management all get simpler, and deployments become reproducible across development, test and production.

Practical consequences:

• Consistent setup across every environment, without manual installation steps

• Straightforward integration with Kubernetes and other container platforms

• Faster onboarding for new environments and proof-of-concept installations

• The same platform across on-premises, private cloud and hybrid infrastructure

This matters most in regulated industries, where deployment control, security and governance are not negotiable. Containerization gives platform teams a clean, auditable way to run digna inside their own perimeter.

QueryMode: choose how your SQL is executed

digna executes its checks inside your database engine. Release 2026.06 gives you control over how those queries are shaped, through a new QueryMode setting with two options.

Single mode calculates each statistic with its own dedicated SQL query. This produces a higher query count, but a much lower memory footprint per query. It is the right choice for very large datasources, or for systems where combined queries risk hitting spool limits or running out of memory.

Combined mode computes all statistics within a single SQL query. Fewer queries, less network overhead, and better throughput when the datasource fits comfortably in memory. This suits frequent, parallel executions on manageable data volumes.

There is no universally correct answer here, which is exactly why it is now a setting. QueryMode lets you tune the balance between performance, resource usage and memory safety per datasource, instead of accepting one execution strategy everywhere.

Configurable prediction model

Anomaly detection in digna is driven by a model that learns the normal behaviour of each time series and flags deviations from it. In 2026.06, that model itself becomes configurable through seven parameters:

• Break Sensitivity

• Outlier Sensitivity

• Memory

• Ridge Strength

• Gap Tolerance

• Outlier Correction

• Plausible Range Tightness

These sit alongside the existing Sensitivity and Memory settings on the tolerance band, and they address a different class of problem. Where the tolerance band controls how wide the acceptable range is, these parameters control how the underlying prediction is fitted in the first place: how quickly it adapts to structural breaks, how much weight it gives to history, how it treats gaps and outliers.

The defaults work well for the large majority of series, and every parameter can be restored to its default at any time. For the series that behave unusually, the seasonal ones, the ones with known step changes, the ones with sparse history, you now have a way to make the model reflect reality rather than working around it. If you are unsure which parameter to reach for, talk to us; we would rather help you tune it properly than have you guess.

A redesigned dashboard

The dashboard has been rebuilt with a modernized UI and a clearer structure. Navigation is more direct, monitoring results and quality insights are easier to scan, and alerts, statistics and charts are more readable.

The goal is simple: less time spent finding the relevant information, more time spent acting on it. Teams checking digna every morning should notice the difference immediately.

Extended import and export for validation rules

Validation rules represent real institutional knowledge. Release 2026.06 makes them far more portable.

Import and export functionality has been extended so that rule sets can move cleanly between environments and projects. You can build and test a standardized rule set in development, promote it to production, and reuse it across teams without rebuilding it by hand. Combined with the SDK, this gives governance teams a proper lifecycle for validation logic: authored once, versioned, reviewed and deployed consistently.

CLI updates

The command line interface has been updated alongside the release:

• SDK integration support

• Improved import and export workflows

• General stability and performance improvements

Who benefits from this release

• Data engineers get automation, SDK access and pipeline integration

• Platform teams get simplified, containerized deployment

• Data governance teams get portable, reusable validation rule management

• Analytics teams get better usability and clearer visibility of results

• Data scientists get programmatic access to profiling and anomaly detection output

Getting started

Release 2026.06 continues digna's strategy of extending observability beyond traditional monitoring and into the broader engineering and analytics workflow. The platform's modular architecture now spans digna Data Anomalies, digna Data Analytics, digna Data Validation, digna Timeliness and digna Schema Tracker, all running inside your own infrastructure.

Existing customers can upgrade through the usual process; upgrades typically complete in under an hour. Full details are in the Release 2026.06 changelog, and the Python SDK is available now via 

pip install digna-sdk.

For a walkthrough of everything in this release, watch the video:

digna Release 2026.06 on YouTube

Questions about the SDK, Docker deployment or prediction model tuning? Get in touch and we will walk you through it.