Microsoft expanded its partnership with Manulife on July 22, adding the Frontier Suite to give the insurer a governed foundation for AI across its global operations.1 Snowflake followed with CoCo, an agentic control plane already used by Fanatics, Thomson Reuters and WHOOP to manage data, models and apps in one governed environment.2
The rollouts extend well past these two vendors. Microsoft and Mistral have expanded infrastructure ties and deepened Foundry and Copilot Studio integration. Box added new security controls for AI agents. Siemens and NVIDIA built self-verifying agentic AI for electronic design automation. Each move pushes autonomous agents deeper into core enterprise workflows, from insurance operations to chip design.
Survey data point to a gap behind the momentum. Enterprise leaders say they lack a clear roadmap and ownership model for agentic AI, even as deployment speeds up.3 Security and privacy concerns remain the top obstacle to adoption, ahead of cost and talent shortages.4
Manulife framed its expanded Microsoft deal around that exact risk. Shamus Weiland said the Frontier Suite gives the insurer "the scale, security, stability and intelligence" needed to "embed AI responsibly across the franchise."1 Snowflake made a similar pitch with CoCo, describing it as a "unified, governed environment" for AI workflows rather than another disconnected tool.2
Sentiment across the announcements is mixed but improving: vendors are building governance into new products instead of treating it as an afterthought, directly answering the top complaint enterprise leaders raise.34
But a control plane or a security feature only works once an organization decides who owns it. Survey data show most enterprises have not made that call yet, even as they sign expanded contracts and launch new agentic platforms.3
The result is a rollout running ahead of its own guardrails: adoption measured in new partnerships and product launches, governance still measured in open questions about ownership and risk.