What AI Engine does Vision AI utilizes

Hi there,

Tosca’s Vision AI engine uses patented neural network technology that has been trained to recognize UI elements on a variety of applications. We launched Vision AI back in 2019 and it is a technology that our machine learning scientists and AI experts built from the ground up. Be not mistaken - it is not a generative AI system, nor is it relying on any existing LLM systems. Again, to emphasize, it is a deep learning technology that we built in-house. Vision AI recognizes various types of objects from buttons, drop downs,  input fields, tables, tab bars and so on, across a variety of different ERP and CRM systems. In a nutshell: machine learning involves training a system using large sets of labeled data, where it is rewarded for correct outputs and punished for incorrect ones to improve its accuracy. This is a different to how Gen AI systems work.

Organizations can use Vision AI to gain benefits like automating really old or new applications that standard engines in Tosca cannot recognize and require customization for. It can perform self-healing, in addition to steering applications like Citrix. It can even create automation for Figma files, PDF files, or any mockups where an application has been sketched on.

Importantly, Vision AI is not trained on customer data. For more information on its data security, you can read here: https://vision-ai.app.tricentis.com/documentation/opsinfo#Data-Security

For more information on how it works in Tosca, read documentation here: here: https://documentation.tricentis.com/tosca/2320/en/content/vision_ai/vision_ai_introduction.htm

Or watch this general webinar about our AI approach: