MWS AI, a division of MTS Web Services, has updated its MWS AI Agents Platform. The platform now allows AI agents to execute self-written code within an isolated sandbox, operate on a schedule, and maintain long-term memory of user interactions.
The sandbox environment provides dedicated computing resources and file access, ensuring that agent-generated code runs securely without compromising internal company systems. It prevents unauthorized access to passwords, keys, and external networks.
A new interface builder enables agents to display custom screens with buttons, forms, and charts instead of standard chat windows. This feature is currently in early access, with a full commercial release scheduled by the end of 2026.
Agents can now autonomously access knowledge bases, such as Confluence pages, by formulating their own queries. Additionally, the platform automates quality control by comparing agent responses against predefined benchmarks, replacing manual verification.
The platform supports the creation of virtual employee teams where multiple agents collaborate to complete end-to-end tasks. Users can define specific goals and requirements, allowing agents to plan, execute, and verify work independently.
