Aurora Mobile Limited (NASDAQ: JG) announced upgrades to its GPTBots.ai enterprise AI agent platform focused on knowledge base reconstruction, workflow execution, and enterprise governance capabilities.

The company said the upgrade addresses limitations in AI agent adoption where agents can conduct conversations but cannot integrate with business systems for production use. The platform now includes knowledge graphs with hybrid vector-and-graph retrieval mechanisms, allowing agents to understand business context rather than performing basic keyword searches.

The upgrade introduces agent-driven form collection that integrates with Aurora Mobile's EngageLab LiveDesk Widget across 14 channels including WhatsApp, Slack, Teams, WeChat, and DingTalk. The new Agent Loop Engine enables multi-turn autonomous reasoning, while Agent-to-Agent protocols allow complex tasks to be divided among specialized sub-agents.

"Enterprises worldwide are shifting from buying tools to buying outcomes," said Chris Lo, Founder and CEO of GPTBots.ai. "Customers don't need another toolbox. They need a solution that understands their business, embeds into their workflows, and has their back when something goes wrong."

The platform includes runtime security features, audit logs, and safety guardrails for production deployment. The upgrade also deepens integration between GPTBots.ai and EngageLab, Aurora Mobile's customer engagement platform.

GPTBots.ai offers two modules - Devspace for building and Workspace for running AI agents across customer operations, knowledge management, and business process automation. Aurora Mobile, founded in 2011, provides customer engagement and marketing technology services.