Sponsored

Stephen Brobst – Chief Technology Officer Ab initio Software

Agentic AI is in your future. Every industry and nearly every individual on planet Earth will be impacted by this technology. Today’s situation is not unlike the mainstream introduction of the internet 30+ years ago. Most people underestimated the impact of the internet when it first emerged into mainstream use in the 1990s after the advent of the World Wide Web in 1991. My prediction is that Agentic AI will be even more impactful.

Agentic AI has superpowers that mirror unique capabilities of its human creators: adaptability, use of tools, learning, and specialisation of skills. AI evolved to Agentic in three distinct steps: (1) prediction and rule-based expert systems for action-taking; (2) generative AI and statistical learning for content creation; and (3) the emergence of Agentic AI with contextual adaptation leveraging its superpowers.

Full-credit Agentic AI goes beyond AI Assistants and Robotic Process Automation (RPA) to deliver autonomous decision-making enabled with planning and reasoning capabilities to facilitate task completion. AI Agents can also work in teams, with different agents bringing specialised skills to the table for collaborative problem-solving. AI Agents collaborate with human subject matter experts to accelerate problem-solving.

Agentic AI, when used effectively, creates new business possibilities. Massive value comes when AI transcends the simple productivity improvements of co-pilots to the enablement of new paradigms. Let’s take financial advising in retail banking as an example. Historically, financial advising is only available to high-net-worth individuals (HNWIs) because the economics of compensating highly skilled financial advisors means that only clients with lots of money can qualify for these services. Agentic AI is a gamechanger in this scenario.

It is now possible to create a financial advising team of AI Agents with specialised skills in areas such as identification of financial goals and timelines, identification of existing financial resources, calculation of monthly and yearly cash flows, assessment of investment sophistication and risk appetite, and, ultimately, the creation of a financial plan. While there is still a professional financial advisor as a human-in-the-loop for evaluating and communicating the financial plan, the team of AI Agents accomplishes 95+% of the work, including natural language conversations with banking clients. Leverage of AI Agents presents the opportunity to democratise the availability of financial advising to a more economically diverse constituency. Human subject matter experts are still in control, but a factor of 20 in throughput improvement for advisors means a much deeper reach for satisfying societal needs in this area.

A critical success factor in successful deployment of AI Agents is AI-ready data. The existence of AI-ready data brings intelligence to Agentic AI. Without data, AI is simply artificial. Good decisions require robust data with strong governance. To create a successful financial plan, detailed data is required to quantify financial goals and the means to achieve them. Additional data is required to assess financial risk and the appropriate methods to mitigate such risk.

A subtle nuance to having AI-ready data is the existence of AI-ready metadata. A knowledge fabric is required to describe the semantic relationship between business concepts and the underlying technical instantiation of data. In addition, an ontology that catalogs the relationships within and across datasets is essential to allow an AI Agent to navigate interrelationships in complex information ecosystems. A full-bodied knowledge fabric also contains information about data quality, data recency, data lineage, and data meaning, which are critical for guiding the right choices on which data to use in which contexts.

Success in Agentic AI necessitates access to detailed data for accurate decisioning and robust metadata for context. Data is located across multiple clouds and on-prem infrastructure, often involving many different database and file system technologies. Data comes in unstructured, semi-structured, and structured formats.

In a modern, AI-enabling architecture, data access is handled by specialised agentic data services. These agents retrieve and prepare data, ensuring that it is secure and ready for use by business transformation agents. This approach allows business transformation logic to be constructed without needing to understand underlying data complexity. This architectural approach also future-proofs the implementation of business transformation agents from the inevitable technology shifts when migrating from on-prem to the cloud and/or from one database platform to another. Modern data platforms, such as Ab Initio, support this framework by providing trusted, governed data access; integration and federation of data across environments; and high-performance and cost-efficient data delivery. These capabilities enable AI to operate at scale with extreme reliably.

The internet redefined how businesses connect and operate. Agentic AI will redefine how they prepare, decide and act. Organisations will fall into three categories: AI-native, those that successfully transform, and those that become extinct.

There are no other choices. The opportunity is significant, but so is the gap between those who are prepared and those who are not. The winners will not be those experimenting with edge use cases or incremental improvements with co-pilots. Winners will be the ones building the AI, knowledge, and compute fabrics that allow agents to work securely, reliably, and at scale.