Agentic AI – Opportunities & Risks

On March 27, 2025, NextWave, in partnership with AI Risk and Appian, hosted "Agentic AI – Opportunities & Risks", an evening dedicated to bringing together industry leaders to discuss how Agentic AI is reshaping the financial services (FS) industry and beyond.

The event has received fantastic feedback, and we are delighted to have shared our Agentic AI models, demos, and real-world examples in such an impactful way.

From automation to risk management, read on to recap the key insights from the presentations. You can also click here to download the slides and watch the recording. 

1. The rise of Agentic AI

In contrast to traditional Generative AI, which creates content in response to prompts, Agentic AI operates autonomously, making decisions, executing tasks, and continuously learning. This capability unlocks operational efficiencies across financial services, reducing manual intervention and streamlining processes.

What sets Agentic AI apart?

• Autonomous – Works independently without continuous human input.
• Adaptive – Learns from past actions and improves over time.
• Outcome-driven – Designed to achieve business objectives, not just assist with tasks.

2. AI maturity in financial services

Financial services firms are at different stages of AI adoption, moving from exploration to optimisation. The six stages of AI maturity include:

1. Exploration – Early experimentation with small-scale AI solutions.
2. Adoption – Targeted AI applications in specific business functions.
3. Expansion – AI solutions deployed across multiple business units.
4. Integration – AI-driven decision-making becomes a core business function.
5. Optimisation – AI is fully embedded into the operating model.

• Key enablers of AI adoption include:
• Leadership & Strategy – Executive sponsorship and clear AI roadmaps.
• Data Readiness – High-quality, well-integrated data sources.
• Governance & Compliance – Ethical AI policies and risk controls.
• Talent & Culture – Upskilling employees for AI-driven workflows.

3. Building real AI systems

The event highlighted key principles for successfully developing and deploying AI in financial services:

• Specific agents outperform generic ones every time.
• Task-focused AI agents deliver better performance than general-purpose models.
  
  
• Agentic prompt engineering is less about keywords and more about designing personas.
• AI effectiveness improves when prompts define roles, behaviours, and objectives rather than relying on static keywords.
  
  
• How your agents talk to each other matters more than you think.
• Multi-agent systems require effective communication protocols to ensure seamless collaboration and decision-making.
  
  
• Choosing the right LLM isn’t just technical—it’s an economic decision too.
• Model selection impacts costs, scalability, and operational efficiency—not just accuracy.

  ---

4. Transforming financial services with Agentic AI

The event highlighted several real-world applications of Agentic AI in financial services, including:

Collateral Agent Automation

• Replacing manual email monitoring and spreadsheet-based workflows with automated AI-driven solutions.
• Cutting process times from days to minutes, improving accuracy.

Product Control Agents

• AI-powered digital workers handling front-office data sourcing, reducing operational risk and manual effort.
• Enhanced real-time financial reporting and compliance checks.

Regulatory Compliance & Risk Management

• AI-driven risk assessments for proactive fraud detection and real-time compliance tracking.
• Automated regulatory reporting, reducing costs and improving accuracy.

5. Opportunities & risks of Agentic AI

Opportunities

✔ Process Optimisation – Automating complex workflows to increase efficiency and reduce errors.
✔ Scalability – AI-driven operations that adapt and grow with business needs.
✔ Personalisation – Enhancing customer experiences through hyper-personalised services.

Risks & Challenges

❗ Ethical Considerations – Ensuring transparency, bias mitigation, and responsible AI governance.
❗ Regulatory Compliance – Navigating evolving AI-related legal frameworks.
❗ Operational Risks – Managing AI reliability, security vulnerabilities, and system failures.