ai-computer-vision
2022.4
false
  • Release notes
    • 2022.4.2
  • Overview
  • Setup and configuration
  • Data storage
    • Data storage
AI Computer Vision User Guide
Automation CloudAutomation Cloud Public SectorAutomation SuiteStandalone
Last updated Jun 27, 2024

Data storage

UiPath values data privacy and aims towards protecting all users' right to privacy when sharing data while using the UiPath Computer Vision AI model.

While using the Computer Vision, data is being stored to:

  • guarantee automation resilience by checking for backwards compatibility when upgrading the AI model
  • improve the AI model
  • respond in a timely manner to customer support requests

In the table below you can find a side-by-side comparison of our current Computer Vision deployment options in regards to data storage.

 UiPath cloud serverOn-premises serverLocal server
AI model update resilience99%90% 1100%
AI data tailoringbespoke data tailoringgeneric data fit 2generic data fit
Mock design-time data storageavailablenot availablenot available
Runtime data storagenot availablenot availablenot available
Full data residency controlnot availableavailableavailable

1 90-99% if AI Learning is enabled

2 bespoke data tailoring if AI Learning is enabled

More details for each Computer Vision deployment options can be found in the next sections.

No matter the deployment option, we highly recommend that development (design time) should be done on mock data. Since the AI model only learns from the shapes and colors of the UI elements and not the real data content, you can use mock data when building any Computer Vision workflow, thus avoiding sending any sensitive information such as personal or confidential data.

Cloud

Important: Production (runtime/robot) data is never stored.

We only store data from our cloud server and only design-time (UiPath Studio) screenshots may be stored to improve the model. These screenshots of automated interfaces are processed on our cloud servers, hosted in Azure. As an exception to this rule, user-reported issues are stored for the purpose of fixing specific problems.

Cloud mock design-time data is learned by default, thus offering a bespoke data tailoring service by shaping the AI model to your specific data, while avoiding overfitting. In this way, automations are resilient to AI model updates.

As a consequence, the AI model improves its generalization power as well (what we call "the generic data fit"), making this a win-win situation both for you and our product.

On-premises/on-premises private cloud

Important: Production (runtime/robot) data is never stored.

For closed networks, no data leaves the company premises.

For open networks, no data leaves the company premises, except for the submitted issues, performed by the user to fix a client specific problem.

On-premises design-time data is not shared. Since we don't have access to this data, the AI model can’t learn from it and it can only detect by generalizing; in other words, we can only offer the generic data fit.

Therefore, we can't ensure 100% backwards compatibility. For instance, a new iteration of the AI model might not detect some UI elements its previous iteration was detecting, possibly failing an automation.

You can access the bespoke data tailoring service by agreeing to have your design-time data included in the AI model’s training set by enabling the Include in AI Learning feature from the CV Screen Scope activity. The only prerequisite is using mock data.



We also offer full data residency control, as the feature can be enabled per scope, on a granular level, while being disabled by default. You have full control over which automation you want covered by this feature.

Local Server

Important: Production (runtime/robot) data is never stored.

For closed networks, no data leaves the company premises because all the processing is done locally on the machine CPU.

For open networks, no data leaves the company premises because all the processing is done locally on the machine CPU, except for the submitted issues, performed by the user to fix a client specific problem.

References

UiPath Data Residency policy is the place where you can find all regions where UiPath Automation Cloud™ and our cloud services can be hosted, depending on your licensing plan and region preference.

The UiPath Privacy Principles for Sub-processors set out the general rules for processing of personal data in accordance with UiPath's instructions, transfers and security of personal data, as well as for cooperation between UiPath and sub-processors.

Security is very important to us. UiPath Security information is available for you to check at any time.

For more information about how UiPath handles and processes personal data, we encourage you to read our Privacy Policy.

  • Cloud
  • On-premises/on-premises private cloud
  • Local Server
  • References

Was this page helpful?

Get The Help You Need
Learning RPA - Automation Courses
UiPath Community Forum
Uipath Logo White
Trust and Security
© 2005-2024 UiPath. All rights reserved.