Manulife (NYSE: MFC) partners with Adaptive ML to scale reinforcement learning across AI platform

Manulife Financial Corporation (NYSE: MFC) has entered into a multi-year partnership with Adaptive ML to embed reinforcement learning as a core operational layer within its enterprise AI platform. The agreement signals a shift away from generalized large language models and toward proprietary, real-time model optimization tailored to regulated financial services use cases.

The collaboration marks a strategic escalation in Manulife Financial Corporation’s AI infrastructure, focusing on specialist language models (SLMs) that are fine-tuned using feedback-driven reinforcement learning. Adaptive ML, based in New York, will provide its Adaptive Engine as the reinforcement learning core that continuously refines Manulife’s AI agents in production, with emphasis on use cases such as underwriting automation, process execution, and sales decision support.

According to Manulife Financial Corporation’s Global Chief AI Officer Jodie Wallis, the goal is not only to increase AI deployment velocity, but to retain control over operational risk and customization. By combining specialist models with a learning feedback loop, Manulife Financial Corporation expects to improve both the accuracy and cost-efficiency of its AI services in real-world applications.

Why is Manulife moving away from generalist large language models toward SLMs and reinforcement learning?

Manulife Financial Corporation’s decision to build out reinforcement learning capabilities reflects a broader institutional shift across regulated industries where generalist large language models are increasingly viewed as insufficient for scalable, real-time business operations. While foundational models such as GPT and Claude can power early experimentation and pilot deployments, they fall short in domains requiring domain-specific reliability, traceability, and policy-bound output generation.

Adaptive ML’s Adaptive Engine addresses these shortcomings by enabling closed-loop optimization of small language models deployed across internal systems. These models are not only lighter in size and cost to run, but can be fine-tuned continuously using reinforcement learning to align with specific business objectives, performance thresholds, and compliance requirements. Manulife Financial Corporation’s deployment strategy emphasizes precision-tuning rather than model scale, marking a conscious departure from the compute-heavy trajectories pursued by hyperscalers.

For an organization operating in life insurance, wealth management, and asset management across multiple regulatory jurisdictions, the ability to self-govern model behavior in a cost-efficient manner is a competitive differentiator. By investing in reinforcement learning capabilities now, Manulife Financial Corporation is building a longer-term moat around AI ownership, model governance, and responsible deployment.

What are the initial enterprise AI use cases targeted for reinforcement learning optimization?

The first wave of deployments will focus on three domains: automated underwriting quote generation, complex workflow execution, and AI-powered sales advisory. Each of these areas stands to benefit from models that can adapt dynamically to real-time data and user interactions.

In underwriting, reinforcement learning can enable model outputs to reflect evolving risk factors, policy changes, and customer profiles, reducing friction in quote generation while maintaining actuarial integrity. For process automation, the system allows decision agents to learn from edge cases and escalate exceptions more effectively over time. In sales advisory, feedback-driven tuning helps models better predict customer needs, improve product match accuracy, and provide timely nudges for relationship managers.

The ability to continuously optimize model behavior in production—not just in development—is a core capability being prioritized. Manulife Financial Corporation’s investment is aimed at shrinking the model deployment cycle while increasing outcome reliability, a dual imperative for financial services firms navigating both efficiency pressures and regulatory expectations.

How does reinforcement learning contribute to responsible AI and sustainability goals?

Manulife Financial Corporation has explicitly linked its reinforcement learning strategy to responsible AI commitments. Smaller, specialized models that are continuously tuned using real-world feedback typically require less compute than retraining large models from scratch or using static LLM APIs. Over time, this approach can significantly reduce energy consumption per inference or decision.

Moreover, reinforcement learning supports higher degrees of model transparency and control, key elements in the Responsible AI Principles publicly adopted by Manulife Financial Corporation. By fine-tuning models in closed-loop enterprise environments, the company is better positioned to track how outputs evolve, why decisions are made, and what impact those changes have on customers. This is particularly important in financial services, where explainability is not just a governance preference but often a regulatory requirement.

The partnership with Adaptive ML gives Manulife Financial Corporation the infrastructure to experiment rapidly while maintaining oversight—a balance few organizations have fully achieved at enterprise scale.

How might this partnership shape enterprise AI deployment across the financial services sector?

The deal with Adaptive ML reinforces Manulife Financial Corporation’s positioning as an early mover in reinforcement learning deployment within a highly regulated sector. While generative AI has attracted widespread attention, few financial institutions have crossed the chasm from experimentation to production deployment with internal control over learning loops.

If successful, the Manulife–Adaptive ML partnership could become a blueprint for peer institutions seeking to deploy smaller, self-tuning models that are more explainable and operationally efficient. Insurance and wealth management firms in particular may follow suit, shifting their AI strategies from generic model consumption to proprietary fine-tuning capabilities tailored to high-stakes use cases.

For Adaptive ML, the deal serves as institutional validation of its reinforcement learning infrastructure and could position the startup to scale into adjacent regulated industries such as healthcare and government. The inclusion of Adaptive Engine as a learning operations layer also suggests a future where RLOps (Reinforcement Learning Operations) could become as integral as MLOps for enterprise AI systems.

How are investors and institutional stakeholders likely to interpret this AI infrastructure move?

Manulife Financial Corporation’s inclusion in the inaugural Evident AI Index as the top-ranked life insurance firm for AI maturity reflects growing recognition of its enterprise AI ambition. The Adaptive ML partnership underscores that this ambition is backed by long-horizon infrastructure investments rather than surface-level experimentation.

Institutional stakeholders focused on digital transformation, operating leverage, and regulatory compliance may view this development as a signal that Manulife Financial Corporation is strategically future-proofing its AI stack. While reinforcement learning has long been associated with academic and robotics contexts, its entrance into enterprise financial operations marks a notable shift in market readiness.

Manulife Financial Corporation’s capital discipline and execution track record will determine how this AI investment translates into operational and financial impact. But the decision to own and optimize smaller models rather than rent compute-heavy LLMs aligns with a broader institutional trend favoring AI control, transparency, and energy efficiency over brute force scale.

What does the Adaptive ML partnership reveal about Manulife’s enterprise AI strategy and industry positioning?

• Manulife Financial Corporation is embedding Adaptive ML’s reinforcement learning engine into its AI platform to fine-tune small, specialist language models in real time.

• The move positions Manulife Financial Corporation to reduce costs and improve accuracy in core use cases like underwriting, process automation, and sales advisory.

• Adaptive ML’s reinforcement learning operations layer allows for continuous model optimization based on real-world data and feedback, improving alignment with business objectives.

• Manulife Financial Corporation is emphasizing internal control, model governance, and energy efficiency over reliance on general-purpose large language models.

• This signals a growing trend among regulated enterprises toward proprietary model development and fine-tuning instead of outsourced AI capabilities.

• Reinforcement learning supports Manulife Financial Corporation’s responsible AI and sustainability commitments by reducing energy-intensive retraining cycles.

• Institutional investors may see the move as a disciplined, forward-looking investment in AI infrastructure rather than hype-driven experimentation.

• Adaptive ML gains a high-credibility deployment partner in financial services, potentially opening doors to other regulated sectors seeking RL-based enterprise AI infrastructure.