> For the complete documentation index, see [llms.txt](https://documentation.ardoq.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://documentation.ardoq.com/changelog/readme.md).

# Product Updates

New updates and improvements to the Ardoq platform.

{% updates format="full" %}
{% update date="2026-11-02" tags="feature,open-beta" %}

## Build Your Own Agents, now in Open Beta

<div data-with-frame="true"><figure><img src="https://2895392706-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FUSBYkUZEeQBauw5V4iub%2Fuploads%2Fumx0CV4D0IsK2NY77IXq%2Fimage.png?alt=media&amp;token=7403e6e1-28ec-4660-b7d4-0bc0cadb5547" alt=""><figcaption></figcaption></figure></div>

Custom Agents are in open beta. Try them on your own architecture and tell us what to improve.

The best EA analysis often lives in someone's head, or in a prompt you retype every week.

**Custom Agents** let you write that analysis once, as step-by-step instructions, then run it from the AI Assistant on your own architecture.

What you can do in the open beta:

* **Capture your method once.** Define the role, the steps and the output you expect. Add <\<variables>> so one agent works across workspaces and component types.
* **Give it only the tools it needs.** Choose toolsets such as Data Population, Document Processing or ShiftX Processes. Core access to your Metamodel, Viewpoints, Reports and Dashboards is always included.
* **Everyone Can Run it in one click.** Press play on an agent card, or ask the Assistant to run the Stakeholder Map agent.
* **Stay in control.** Agents run with the permissions of the person using them, so nobody sees more than they should, and changes land as draft Scenarios for review.

Beta means we want your input. Pick one task your team repeats, build an agent for it and run it a few times. Then tell us where the instructions, tools or results fell short.

Your practice already has methods that work. Now they can run on demand, for everyone on your team.

<a href="https://help.ardoq.com/en/articles/866857-build-your-own-custom-agents" class="button primary">Try Custom Agents</a>
{% endupdate %}

{% update date="2026-10-26" %}

## Coming 2 November: Create References Without Leaving Your View&#x20;

On 2 November, we will release an improvement that simplifies creating references and helps you model without leaving your view.

### What’s changing

Creating references should not interrupt your flow. Previously, adding a reference opened the Component Overview Page over your diagram, making it harder to keep the visual context of what you were connecting.

From 2 November, you can create references in a side panel that opens alongside your diagram, so you can stay in context while choosing the right component and reference type.

<div data-with-frame="true"><figure><img src="https://2895392706-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FUSBYkUZEeQBauw5V4iub%2Fuploads%2FYYfL35DtphqBey1UuvJ0%2Freferences-nov-2026.gif?alt=media&amp;token=dd2bde97-42c3-4773-a4b0-f67065972344" alt=""><figcaption></figcaption></figure></div>

### Why it’s better

* Keep your place. Your view stays visible while you add references, preserving the spatial context that makes visual modeling intuitive.
* See only what you can use. The panel shows the reference types your metamodel and permissions allow, reducing dead ends and keeping the model consistent with your team’s structure.
* See the full picture. The component’s existing references appear in the same panel, so you know what’s already connected before adding more.

### Thank you for your feedback

This was a highly requested improvement, so thank you for helping us shape it. If you want to try it before 2 November, in-view reference creation is available now in open beta; enable it in Feature Settings.

<a href="https://help.ardoq.com/en/articles/711686-in-view-reference-creation-open-beta" class="button primary">Learn about creating references directly in the view</a>
{% endupdate %}

{% update date="2026-11-02" %}

## Data Provenance: Trace Every Change to its Source&#x20;

<figure><img src="https://2895392706-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FUSBYkUZEeQBauw5V4iub%2Fuploads%2FWmWjdvE1NxbKAScYkjsQ%2Fprovenance-2026.gif?alt=media&amp;token=dbd5b49c-0b9a-49e4-9bba-f17f9a52913e" alt=""><figcaption></figcaption></figure>

Until now, Ardoq told you who last changed a component, not how. An integration sync, a survey answer and an AI Assistant edit all looked the same.&#x20;

Now every new component and reference carries two stamps:

* Origin: how it was created.
* Updated via: what last changed it. An integration, the API, a survey response, an edit in Ardoq or the Ardoq AI Assistant.

What that means in practice

* Before a review, check that the data you present came from your system of record, not a one-off edit.
* When the Ardoq AI Assistant writes data, reviewers see it labeled and know what to check.
* When a value looks wrong, see which integration wrote it and fix it at the source.

Find both fields in Data Inventory, Component Overview page and Audit Log. Data created before this release shows Unknown. Once a scheduled integration sync has run this will update.&#x20;

Data Provenance is the first step in the data quality work we’re building in Ardoq. Knowing where data comes from is what makes it possible to check, improve it and drive more confident decisions.
{% endupdate %}
{% endupdates %}