The biggest topic for companies this year is to apply artificial intelligence (AI) agents to in-hous..
https://www.mk.co.kr/en/it/12009300
Publish Date: 2026-04-06 03:37:00
Source Domain: www.mk.co.kr
사진 확대 John Jester, Chief Revenue Officer, Beam Software
The biggest topic for companies this year is to apply artificial intelligence (AI) agents to in-house systems. This is because employees create their own AI agents to streamline their work, and the agents completely automate repetitive tasks within the company and greatly increase productivity.
At the same time, the biggest concern is how to manage AI agents. Neglected agents act as a channel for data leakage and as a security risk for companies.
In response to this question, beam software chief revenue officer John Jester recently met with Maeil Business Newspaper and emphasized that “if you don’t use AI agents to give them identity and manage access (access), it’s no longer valid.”
Beam Software is a company that develops software that helps companies manage and restore data in various environments, and it acquired SecurityAI last year in line with the rising trend of AI agents, expanding related capabilities.
The difficulty for companies described by Jester CRO is to provide AI agents safely. This is because companies have difficulty monitoring what data in-house agents access and whether they do not work abnormally. It is difficult to even figure out the agents that employees personally produce using Claude. He explained, “When I meet with information protection officers (CISOs) around the world, it is difficult to apply and use functions such as security and access rights to Generative AI.”
To this end, the data software provided by the beam software scans the agents used in-house in real time and then identifies what data they are accessing and what policies are being applied to each agent.
In addition, even if it is unstructured data, the information in it is classified by type such as name, phone number, age, key, and account information, and sensitive data is labeled separately to make it easier to distinguish agents who need…
