People and multiple AI agents collaborate together in a workplace chat room. /Courtesy of Lee Ho-jun

The era of "agent collaboration," in which artificial intelligence (AI) participates like a person to take on or assist with tasks in work processes, is opening.

NHN Dooray held a press briefing on the 28th at its Pangyo office in Seongnam, Gyeonggi, and unveiled the direction for revamping Dooray, its collaboration tool with expanded "AI agent" features. It is a large-scale update about one year and six months after introducing "Dooray AI" with AI features in 2024.

The core of this revamp is that AI has evolved from simply answering to becoming an agent that performs real work. In existing collaboration tools, AI was limited to providing summaries or Q&A based on internal data such as mail, messenger, and documents. But now, it links to external systems to use data and even carries out tasks, NHN Dooray said.

In particular, it has changed so that AI can be used like a member of the organization. Users can search for the agents they need in the messenger as if searching for employees and have one-on-one conversations, and they can invite them to group chats. Unlike chatbots, it is also possible to deploy multiple agents in a single chat room to divide up the work.

Agent functions are divided by use. "My agent," which supports personal work, operates based on personal data and work history such as mail, calendar, wiki, and drive. For example, when applying for leave, if you ask the agent for the approver and schedule, it will automatically complete the approval. This used to require going back and forth between the calendar and the approval system.

In collaboration processes, a "project agent" is used. It operates based on task, document, and schedule data within the project and, beyond simple queries, also has execution permissions such as sending messages. For example, if you need to share related content with another department after a meeting, the agent summarizes the minutes and delivers them to the necessary staff. If a project is running later than expected, it can also leave comments for the task owner and perform similar actions.

Baek Chang-yeol, CEO of NHN Dooray, said it is "not just a chatbot that operates based on project data, but an agent with data 'ownership.'"

At a press briefing at the Pangyo office in Seongnam, Gyeonggi Province, on the 28th, Baek Chang-yeol, CEO of NHN Dooray, introduces the features. /Courtesy of Lee Ho-jun

It also offers expert-level agents that can be used immediately without separate development. These agents are specialized AI assistants that have completed training on data in professional domains such as legal, accounting, regulation, and security. For example, if you request disclosure information of a specific corporation, it automatically compiles the relevant details and analysis results into a document. It can also produce the analysis results as documents or tables in PDF or HWPX formats.

NHN Dooray plans to move away from its existing flat-rate plan and introduce a pay-as-you-go plan in which the expense varies with AI usage. While the base fee will be lower, costs will increase the more you use features such as document generation or data analysis. Major overseas providers such as Google Cloud also use pay-as-you-go pricing.

Meanwhile, NHN Dooray plans to accelerate market expansion, focusing on the public and financial institutional sectors. In the public market, it will try to move beyond the existing internet network into the administrative network. Starting with the administrative network of the Daegu National Information Resources Service, it plans to expand direct sales to the central government and local governments.

In finance, recent regulatory easing is expected to be a variable that expands the market. The Financial Services Commission and the Financial Supervisory Service have allowed cloud-based software as a service (SaaS) to be used on internal business networks in finance by revising the detailed enforcement rules of the Electronic Financial Supervisory Regulations starting on the 20th. As a result, related systems can be introduced without going through the designated innovative financial service process, and the pace of SaaS adoption in finance is expected to accelerate further.

Baek said, "We are focusing on finance and the public sector," adding, "In one year, we secured 25 references and are proceeding with additional contracts with more than 10 clients, including a tier-1 bank."

※ This article has been translated by AI. Share your feedback here.