> For the complete documentation index, see [llms.txt](https://cleyrop.gitbook.io/docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://cleyrop.gitbook.io/docs/documentation-fr-en/support-and-resources/changelog-and-release-notes/release-note-v4.5.md).

# Release Note v4.5

## 📣 Announcements

v4.5 consolidates two major areas: AI flexibility and data governance. The global assistant gives way to the Druid, administrators can now connect their own LLM models, and data quality becomes more expressive with multiple alert thresholds. This version also lays the foundations for better security control, with active session management and new authentication policies. v4.5 also marks an evolution in the platform terminology: the Agent Factory becomes The Factory, Agent Apps and Agent Tools become Apps and Tools, and AI Contexts become Agents.

***

## 🚀 New features

### **Welcome email**

* A welcome email is now sent upon first login. It contains the link to the platform as well as the documentation.
* Existing users will also receive this email at their next login.

### The Factory - Terminology evolution

* The **The Factory of Agents** is now called **The Factory**.
* The **Agent Apps** and **Agent Tools** become **Apps** and **Tools** — the whole being referred to by the term **Apps**.
* The **AI Contexts** become **Agents**.

***

### The Druid

* The global assistant becomes The Druid. The "New Chat" button now opens the new Druid interface.
* User environment variables for Tools can now be configured directly from The Druid: if a Tool requires configuration, it appears grayed out with a "Configure" button in the listing.

→ [The Druid](/docs/documentation-fr-en/le-druide/chat-securise.md)

***

### Models & API Keys

* Platform Managers can connect their own LLM model providers (OpenAI, Mistral, Anthropic, OVH, etc.) from a dedicated interface in Administration.
* Three default models can be defined: instruct (used by the Druid), embedding (used by corpora), and coder. Changing the default model takes effect immediately across the entire platform.
* The list of models available from a provider can be retrieved automatically via the "Discover models" button.
* Keys preconfigured by Cleyrop remain visible and usable, but they cannot be modified or deleted.

→ [Manage models and API keys](https://cleyrop.gitbook.io/docs/administration/gerer-les-modeles-et-cles-api)

***

### AI consumption monitoring

* A dashboard for monitoring token consumption model by model is now available (Administration > Monitoring, Tokens tab).
* For the selected period, it displays: prompt tokens, completion tokens, and total tokens.

→ [Manage models and API keys](https://cleyrop.gitbook.io/docs/administration/gerer-les-modeles-et-cles-api)

***

### Data quality — Multiple alert thresholds & modification of an existing rule

* Data Quality rules are no longer limited to a boolean return. They can now return numeric values (integer, decimal), text, or date.
* For each return type, you can define up to 3 conditions that can be combined with AND / OR, each associated with a severity level (Critical, Major, Minor).
* If no condition is met, no alert is generated.
* It is now possible to modify a rule after its creation: name, description, SQL query, and alert thresholds.
* Changes take effect at the next dataset refresh; they are not retroactive.

→ [Data quality](https://cleyrop.gitbook.io/docs/gouvernance-de-donnees/qualite-des-donnees)

***

### Referencing output datasets in the library

* Datasets produced by a Dataflow are no longer automatically visible in the project library. Only explicitly referenced datasets appear there.
* A button **Reference** is available directly on the Dataflow graph above the output dataset, as well as from its detail page.
* The referenced dataset becomes visible in the library **once the Dataflow is deployed to production.**

{% hint style="info" %}
Datasets already present in the library before the version upgrade are not impacted.
{% endhint %}

→ [Manage Datasets](https://cleyrop.gitbook.io/docs/projet-data-and-ia/datasets/gerer-les-datasets)

***

### Dataset notification management

* The option to subscribe to refresh failures is now explicitly offered when creating a dataset. The owner of a dataset can disable their automatic subscription to Data Quality rules from Notification Management.
* Subscription to alerts is only offered if at least one rule is defined on the dataset.
* When a rule is created, an invitation to subscribe is displayed if the user is not yet subscribed.

→ [Manage Datasets](https://cleyrop.gitbook.io/docs/projet-data-and-ia/datasets/gerer-les-datasets)

***

### Deleting a DevSpace

* Users can now delete their DevSpace from the interface, via a button **Delete** available on the DevSpace details page.
* Deletion is irreversible and results in the permanent loss of all data stored in /home/cleyrop.

→ [DevSpace](https://cleyrop.gitbook.io/docs/gen-ai/devspace)

***

### Active session management

* A new tab **Sessions** is available in Personal Space. It displays the devices currently connected to your account: IP address, browser, start date, and last access.
* Each session can be revoked individually. The current session cannot be closed.
* A button **Revoke all other sessions** allows you to disconnect all other devices in a single action.

→ [Personal Space](https://cleyrop.gitbook.io/docs/espace-personnel/profil-utilisateur)

***

### Retaining ownership when deactivating an account

* When deactivating a user account, two options are offered:
  * **Reassign ownerships** (default option): assets are transferred to the project manager.
  * **Retain ownerships** : assets remain assigned to the deactivated account and are automatically restored to it upon reactivation.

{% hint style="warning" %}
By retaining ownerships, assets will remain assigned to an inactive account. Make sure critical assets remain accessible and manageable during the deactivation period.
{% endhint %}

→ [Manage users and groups](https://cleyrop.gitbook.io/docs/administration/gerer-les-utilisateurs-et-groupes)

***

## Bug fix

* Fixed the display of the details of a partial Dataflow execution.
* Fixed scrolling in the add-user modal.
* Fixed display issues in the Agents journey.
* Fixed the coder model "Update" button, which was incorrectly displayed in a grayed-out state.
* Fixed pagination in Work Data (display limited to 50 items).

***

## Known Issues

* **Display of user status** : user status may display incorrectly on the detail page, with an inconsistent activate/deactivate action. This issue is purely visual and has no functional impact. Fix planned for v4.5.1.