Most financial institutions are still treating AI like a science project — sandboxed, cautious, and kept well away from anything that actually matters. Manulife, one of Canada’s largest insurers, is making a different bet: it’s moving AI agents directly into the operational core of its business, not the margins of it. That shift is more significant than it sounds.
The company is deploying what’s known as agentic AI — systems that don’t just answer questions but actively carry out multi-step tasks across internal software, databases, and workflows. For a firm managing policies, claims, underwriting assessments, and financial reporting at global scale, that’s not a small experiment. It’s a structural change in how work gets done.
What “Agentic AI” Actually Means in Plain Language
Most people have experienced AI as a chatbot — you ask it something, it responds, conversation over. Agentic AI works differently. Think of it less like a Q&A tool and more like a capable intern who doesn’t just answer your question but goes off, pulls the relevant files from three different systems, cross-references them, prepares a summary, and has it ready before your meeting starts.
That’s the practical version of what Manulife is building. Its new runtime platform allows teams to deploy agents that interact with internal systems sequentially — collecting data, processing it, and preparing outputs for human decision-makers. The human stays in control of the final call, but the agent handles the information-gathering groundwork that currently consumes hours of staff time.
This matters because insurance is one of the most data-heavy industries on earth. A single underwriting decision might require pulling structured data from policy records, cross-referencing claims history, checking financial exposure models, and summarizing regulatory notes — all before a person even begins evaluating the case.
The $1 Billion Signal Worth Paying Attention To
Manulife has publicly stated it expects its AI initiatives to generate over US$1 billion in value by 2027, primarily through productivity gains and workflow automation. That’s not a marketing number — it’s a forecast that has to survive scrutiny from investors and regulators alike.
What makes it credible is the baseline they’re working from. The company already has more than 35 generative AI use cases in active production, with plans to scale that to approximately 70. Perhaps more telling: roughly 75% of its global workforce reportedly uses generative AI tools in some capacity today. That’s an unusually high adoption rate for an organization of this size and sector.
The jump from generative AI tools — writing, summarizing, drafting — to agentic AI systems that act, decide, and execute across systems is the critical transition most organizations haven’t made yet. Manulife appears to be making it deliberately and at scale.
Why Most Companies Are Still Stuck at the Pilot Stage
According to McKinsey’s 2024 Global AI Survey, about 65% of organizations now use generative AI in at least one business function — nearly double the figure from the previous year. That sounds impressive until you look closer: only a small fraction of those deployments have reached full production across meaningful parts of the business. Most remain confined to pilot programs or single departments.
The gap between “we’re piloting AI” and “AI is running inside our core operations” is vast. It involves integration with legacy systems, staff retraining, governance frameworks, and — in regulated industries — substantial compliance work. Companies often get stuck in the pilot phase not because AI doesn’t work, but because the organizational and regulatory infrastructure to scale it safely isn’t in place.
Manulife’s decision to build a dedicated runtime platform for agentic AI suggests it has moved past that bottleneck, or is at least investing seriously in doing so. That’s a distinction most financial firms haven’t been able to make convincingly.
The Regulatory Tightrope That Makes Finance AI Hard
Deploying AI inside a bank or insurer isn’t like deploying it at a tech startup. Every system that touches underwriting decisions, risk assessment, or investment analysis must be auditable. Regulators need to be able to trace how a decision was made — which data was used, what logic was applied, and whether the outcome was fair and explainable.
That requirement rules out a lot of AI approaches that work fine in lower-stakes contexts. It also means financial institutions spend heavily on model oversight, internal AI governance policies, and risk review processes before anything goes near a live workflow. Deloitte’s research on AI in financial services confirms this: investment in oversight infrastructure is growing alongside investment in the AI systems themselves.
For Manulife, this means the agentic AI deployment isn’t just a technology project — it’s also a compliance and governance project running in parallel. Getting that right is arguably harder than building the agents themselves.
Quick Reference: Manulife’s AI Deployment at a Glance
| Metric | Detail |
|---|---|
| Expected AI Value by 2027 | Over US$1 billion |
| Current GenAI Use Cases in Production | 35+ |
| Planned Use Cases (Near-Term) | ~70 |
| Workforce Using GenAI Tools | ~75% globally |
| AI Approach | Agentic AI via dedicated runtime platform |
| Primary Focus Areas | Internal operations, underwriting support, decision assistance |
| Industry Context (McKinsey 2024) | 65% of orgs use GenAI, but few at full production scale |
What This Signals for the Broader Finance Industry
When a firm the size of Manulife moves agentic AI into core workflows, it creates pressure on competitors to follow. Other insurers and financial institutions watching this deployment will be measuring it against their own pilot-stage experiments — and asking uncomfortable questions about why they’re still at the proof-of-concept stage.
The broader trend here is what analysts are calling the “operationalization” of AI — the shift from AI as a tool people use occasionally to AI as infrastructure embedded in how an organization functions every day. Financial services, healthcare, and logistics are the industries where this transition is moving fastest, because the volume of structured data and repeatable decision-making creates obvious entry points for automation.
What Manulife is demonstrating is that this transition is achievable inside a heavily regulated, risk-conscious environment. That’s the signal other financial firms will find hardest to ignore.
The Next 12–24 Months: Where This Is Heading
Over the next two years, I expect we’ll see a meaningful split emerge between financial institutions that have successfully operationalized AI and those still running pilots. The gap will show up in processing speeds, cost structures, staff productivity metrics, and — eventually — competitive pricing power in products like insurance premiums or lending rates.
Agentic AI will likely expand from back-office tasks into more complex areas: real-time risk monitoring, regulatory reporting, and eventually elements of customer-facing advisory workflows with appropriate human oversight baked in. The firms that built their governance infrastructure early — as Manulife appears to be doing — will have a meaningful head start.
The companies that waited to see how others handled the regulatory complexity will find themselves playing catch-up in a market where the operational baseline has quietly shifted beneath them.
If you’re tracking how AI is genuinely changing the way large organizations operate — beyond the headlines and the hype — Manulife’s deployment is exactly the kind of case study worth following closely. It’s not flashy. It’s not consumer-facing. But it may be one of the clearest early examples of what enterprise AI actually looks like when it grows up. I’ll be writing more about agentic AI deployments across financial services and healthcare — subscribe to stay ahead of these shifts as they develop.