Stax CTO Claims New Framework Transforms Agentic AI
Understanding the Future of Artificial Intelligence in Business
The impact of artificial intelligence (AI) on our daily lives and business operations is profound. This technology has the potential to reshape how we work and interact, but there are significant challenges to overcome in order to fully realize its benefits.
The Cost and Limitations of AI Models
AI models are not only expensive to develop, but they also face scalability issues. Mark Sundt, the Chief Technology Officer at Stax Payments, highlighted these limitations in a recent discussion about the evolving landscape of AI in finance and banking. He noted that many existing models are only superficially effective, answering only a handful of questions. This raises concerns about their ability to provide meaningful insights. Sundt’s observations stem from a discussion in the "What’s Next in Payments" series, which emphasizes the need for AI systems that can communicate and interact coherently.
Moving Towards a Distributed Model
Rather than relying on centralized AI applications with limited functionality, firms are exploring distributed models that enhance data management and workflow efficiency. This shift is crucial for enterprises aiming to streamline operations and reduce costs associated with traditional, monolithic AI systems. As businesses begin to integrate AI more deeply, it is essential for all stakeholders—from development teams to executives—to align on the business needs and expected outcomes of implementing agentic AI.
In November, a significant advancement in this area was introduced by Anthropic with the launch of the Model Context Protocol (MCP). This protocol provides a standardized method for connecting AI agents to internal data, enabling smoother integration and reducing system fragmentation.
Different Approaches to AI Integration
There are two primary approaches to adopting AI technologies:
- Client-side applications: These involve directly using AI models as tools for various tasks.
- Server-side functionalities: This entails exposing capabilities from existing applications that employ large AI models.
Sundt emphasized that a blend of these approaches can create a hierarchical network of models that work in collaboration to optimize business processes.
Practical Applications of AI
One of the notable applications of the MCP is in the area of Know Your Customer (KYC) processes, where it allows real-time assessment of documentation. This ensures data follows the correct pathway for effective decision-making. Sundt has already begun leveraging existing libraries and tools to enhance operations at Stax Payments.
For instance, after acquiring a new company, Sundt utilized agentic AI and MCP to conduct a Payment Card Industry (PCI) audit. By developing a model based on PCI specifications, he was able to quickly identify compliance issues that had not been thoroughly examined. This innovative approach allowed for a more productive audit process.
Traditionally, Sundt would have spent a considerable amount of time coding different business processes, such as evaluating job candidates. However, he states that leveraging AI and MCP significantly improves these processes, yielding better outcomes compared to older, more manual methods.
The Future Landscape of AI Communication
Sundt believes that MCP represents a breakthrough in how AI models communicate and collaborate. According to him, this protocol will facilitate seamless exchanges between different models, allowing them to discover new capabilities in real time. This development promises to revolutionize how organizations implement AI, elevating its role from a static tool to a dynamic partner in business processes.
As the landscape of AI continues to evolve, it opens up numerous possibilities for enhancing operational efficiency, compliance, and overall business performance. Businesses that can effectively harness these technologies will likely find themselves at a competitive advantage in the marketplace.
