ジェネシス、仮想エージェントのタスク完了を推進

Genesys has introduced the Genesys Cloud Agentic Virtual Agent, built using Large Action Models. The goal is simple. Move customer service AI from answering queries to actually completing tasks across enterprise systems.

Self-service has been underperforming for years. According to Gartner research, the success rate currently stands at 22 percent. The main problem exists because current virtual agents, including those that use large language models, were designed to conduct conversations instead of performing tasks. They can respond well but fail when requests require coordination across systems like CRM, billing, or service operations.

Genesys is trying to change that with LAM driven automation. The agent understands customer intent, decides the next step, and executes workflows across front and back office platforms until the request is resolved. The system includes three components which are governance controls and audit trails and built-in guardrails to ensure compliance while tracking actions. Companies can also configure rules and permissions through Genesys Cloud AI Studio.

- 広告

こちらもお読みください： アドバンスト・メディア、スプレッドシートの採点をAIに置き換え

The platform runs on Scaled Cognition’s APT 1 action model and is currently in early enterprise trials. Wider global availability is expected in early 2027, reflecting a broader shift toward AI that focuses on outcomes, not just responses.