Cisco's AI Agents Revolutionize Office Work, But Security Concerns Linger

Cisco, the world's largest network equipment company, has introduced AI agents to about 90,000 employees worldwide. The agent's name is 'MyAgent', which goes beyond just a chatbot and actually performs most of the work of office workers.

The AI agent uses in-house information and work tools accessible to individual employees to perform tasks such as data research, analysis, report writing, and email management. According to Cisco's chief financial officer, Mark Paterson, 'AI agents have already written 80-90% of the financial disclosure drafts.'

However, some analysts say that it will take a considerable amount of time for AI to replace human employees as side effects caused by AI's work area is expanding too rapidly. In fact, Meta's attempt to delegate tasks to AI agents has resulted in increased code changes and new services or improvements delivered to actual users have not seen a significant increase.

The security issues also emerge, with AI agents making mistakes such as inquiring customer information or leaking security files to the outside without permission. Experts advise that high-risk behaviors should be approved by humans, and AI's entire work process should be recorded to clarify the cause and responsibility in the event of an accident.

The gap in AI capabilities between companies that use a lot and those that do not is widening. According to OpenAI, leading companies with the top 10% of monthly AI usage have 8.3 times the number of AI output tokens per active user compared to companies with moderate usage.