Using VALD AI for data analysis

VALD AI gives you the ability to easily query and ingest your data straight from VALD Hub.

In the current Beta phase, its role is limited to data exploration and reporting, not providing interpretations or recommendations.

This feature is currently available as a closed Beta, meaning only a limited set of users have access.

Get access to VALD AI

About the Beta

This release of VALD AI is made available to a restricted group of VALD users via an external vendor (Snowflake) for the purposes of gathering feedback. When using the Beta version, keep in mind that:

• some features may be incomplete, behave unexpectedly, or change over time, and
• you may encounter bugs or limitations that will not be present in the final product.

The VALD team are actively monitoring and resolving issues based on your feedback. We greatly appreciate any time and effort spent exploring VALD AI and sharing what works or doesn't.

Help us improve VALD AI

	Use the thumbs up button when the output matches what you were looking for, and highlight any areas that are working well.
	Use the thumbs down button when the output was not what you expected, and give details about what was missing or incorrect.

Read through our prompting tips and best practices for guidance on getting the most out of VALD AI.

Capabilities and limitations

VALD AI can...

• Securely query your organization's VALD Hub data – en masse or for specific cohorts
• Aggregate and visualize your data in a preferred format (table, chart*, etc.)
  * Radar and quadrant charts not available in the Beta.
• Carry out new calculations from your existing data
• Help you inspect the quality of datasets and identify outliers
• Create normative-style reports comparing data to your own custom benchmarks

VALD AI cannot...

• Provide recommendations or clinical feedback for decision-making
• Interpret your data beyond presenting the queried results, calculations or visualizations
• Infer or compare against data from organizations you don't have access to
• Utilize global (age or sex-based) normative datasets in the Beta release
• Display real-time data – data is refreshed daily, meaning very recent tests may not appear until the next day
• Localize date and time – all dates will be displayed in UTC

See below for a list of available VALD systems from which data can or cannot be queried in the Beta.

VALD system or data	Supported
NordBord test data
ForceFrame test data
HumanTrak test data
ForceDecks test data
DynaMo test data
SmartSpeed test data
Profile data from VALD Hub
ForceDecks tags
MoveHealth data
Repetition-level data
Force trace data
Video or other media files
Other test parameters or notes

This Beta includes a defined, limited scope – broader support is expected in upcoming releases.

Access and data security

For the purposes of the Beta, VALD AI is hosted in a secure Snowflake environment accessible through a dedicated, separate URL.

This requires you to authenticate your access by using the Sign in using VALD option on the log in screen.

VALD AI can only access data from your organization's VALD Hub.

Prompting tips

See our guide on "Prompting best practices for VALD AI".