Seoul-based Infofla has launched Version 2 of its Selto automation platform, targeting persistent challenges with traditional robotic process automation that fails when user interfaces change unexpectedly.

The company claims its "VLAgent" engine combines large language models with visual recognition to enable automated workflows that adapt to interface modifications, pop-ups and layout changes that typically break conventional RPA deployments.

Traditional RPA implementations face significant challenges when applications update their user interfaces, with bots failing to complete tasks when buttons move or interfaces refresh. Industry analysts report that up to 50% of initial RPA projects fail, often due to the technology's inability to adapt to changing environments.

Selto V2 introduces visual confirmation capabilities that allow users to see what the system has learned from screens, according to Infofla CEO In-mook Choi. The platform can now process external data sources and create conditional workflows based on user types or interface conditions.

The South Korean Ministry of the Interior and Safety has deployed Selto. The company reports it is conducting proof-of-concept trials with financial and insurance sector clients.

The platform's ability to handle "unpredictable digital environments" represents a significant technical challenge that traditional RPA vendors are also addressing through AI integration.

Major RPA providers are incorporating generative AI and machine learning to improve bot resilience and decision-making capabilities.

The company offers desktop installation allowing individual users to train automation workflows, potentially appealing to organisations seeking to democratise automation beyond IT departments.

https://www.infofla.com/en/home