Go to contents

Infofla's VLAgent-based 'Selto' gets smarter with the V2 update.

Infofla's VLAgent-based 'Selto' gets smarter with the V2 update.

Posted July. 18, 2025 14:15,   

Updated July. 18, 2025 14:32

- Infofla's VLAgent-based 'Selto' has been updated to Version 2, enhancing 
  its performance and features
- The update introduces visual confirmation of learned screens, conditional 

  branching, and variable features, significantly improving automation precision
  and flexibility.
- Infofla's business automation solution is now being tested in the financial and

  insurance industries

While many businesses are looking to automate their operations with AI, they often face significant hurdles. The main challenges are finding AI solutions tailored to their specific needs and dissatisfaction with the performance of current AI technologies.

'VLAgent', an AI-based workflow automation solution by Infofla, is a noteworthy answer to these challenges. This is powered by a VLM(Vision Language Model) technology that enhances an LLM with image recognition. This enables the AI to directly perceive on-screen elements, learn on its own, and, most distinctively, handle diverse scenarios with human-like proactive responses.


Inofla \

Inofla 'Selto' with enhanced overall functionality through V2 update / source=Infofla


The existing RPA-based automation solutions were limited to mechanically performing a few predefined patterns and were vulnerable to unexpected situations. For example, in automating website management, existing solutions would fail to cope and generate an error if a pop-up window suddenly appeared or the web design changed. However, the Vision AI-based VLAgent can continue its normal operation, just like a human, by closing the pop-up window or recognizing the modified web design.

Infofla has consistently upgraded its VLAgent since its initial launch in mid-last year. Selto, the first business automation service utilizing the VLAgent engine, recently updated a V2 (Version 2), significantly enhancing its overall performance and features.

The basic usage remains the same: users enter a task prompt, and detailed tasks are generated and processed automatically. However, this update promises smarter and more flexible business automation.

The most notable update is the ability to visually confirm the screens that AI has learned. This feature, located in the task details screen, allows users to easily diagnose and correct issues if the automation doesn't work as expected. This is anticipated to improve task reliability by quickly identifying the causes of various errors.

The Vision AI agent on Selto can cope with various unexpected situations/ source=Infofla

The Vision AI agent on Selto can cope with various unexpected situations/ source=Infofla


An 'if (conditional branching)' feature has been added, and automation proceeds differently depending on various conditions. The VLAgent engine's core str
ength has always been its ability to proactively adapt to diverse variables, and this update significantly enhances that capability.

This means adjusting the workflow based on different conditions. For example, even on the same screen, the button customer press might vary depending on the customer type.

Infofla has also added a 'variable' feature that allows to store frequently repeated information like names, addresses, and numbers as variables for use in automation. For example, if you need to automatically input a customer's name, you can set the name as a variable for easy repeated entry. This also allows you to directly apply diverse data received from external systems to your automation.

AI can recognize the screen directly and automate various tasks / source=Infofla

AI can recognize the screen directly and automate various tasks / source=Infofla


New updates enhance automation precision for complex tasks through 'if' and variable features, significantly reducing human intervention. This also increases the practical utility of automation by allowing the same task to be flexibly configured for various situations.

With the Selto V2 update, Infofla has also included an administrator dashboard for easy monitoring of automation progress. Furthermore, they're providing a high-performance server environment for large-scale testing, boasting two A6000 GPUs or one A100 GPU. This means customers evaluating Selto can now perform their PoC(Proof of Concept) in a significantly more stable and efficient setting.

Infofla's business automation solution has been validated by its implementation in a central government ministry and a well-known shopping mall. Based on this success, Infofla is now conducting PoCs for Selto V2 within the financial and insurance industries.

In-mook Choi, CEO of Infofla, stated, "The core of this V2 update is enhanced intuitiveness and flexibility, allowing anyone to easily design and utilize automation by actively reflecting customer feedback. We will continue to advance AI automation features that maximize work efficiency."


By Young-woo Kim (pengo@itdonga.com)