Manulife expanded its partnership with Microsoft, adopting the Microsoft Frontier Suite to advance AI across its global insurance operations.1 Chief digital officer Shamus Weiland called it "a critical enabler" of Manulife's shift into "a truly AI-driven organization."1
The suite gives Manulife "the scale, security, stability and intelligence" to embed AI "responsibly across the franchise," Weiland said.1 It's the latest sign insurers are pushing agentic AI past pilots and into core operations.
Enterprise content platform Box is building the guardrails that kind of deal requires. The company unveiled administrative controls letting clients deploy AI agents without losing oversight of sensitive data.2 Client Nomura Research Institute said the controls span "prevention, detection, and response" across AI agent access to content.2
Nomura executive Tatsutoshi Murata said Box's multi-vendor support lets the firm "flexibly switch between AI models" as it scales agent use, calling that flexibility "extremely reassuring."2
The operational pain driving this buildout shows up at the startup level. Casap, a dispute-resolution automation company, was founded by Shanti and Sayisi after both encountered the same problem at separate large fintech firms: resolving customer disputes required heavy internal engineering effort even with strong teams.3 Investors backed the company partly because the founders had lived the problem firsthand.3
These moves fit a broader enterprise pattern: platform partnerships like Siemens-NVIDIA and Mistral's integration into Microsoft's Copilot Studio are pairing with AI-native startups targeting narrow verticals, from healthcare admin to business-process automation.
But according to the Agentic Enterprise Report 2026, that pace is the problem. Large majorities of executives deploying agentic AI report no clear ownership model and no pre-deployment governance roadmap. Box's controls, Manulife's "responsible AI" framing, and Casap's dispute-automation pitch are all responses to the same gap: agentic AI is already inside workflows that touch money and customer data, while the structures meant to govern it are still being built after the fact.
