How to win with AI: focus on business outcomes

The popularity of AI in 2023 is undeniable. From generative art, music, and movies to sophisticated chatbots seemingly capable of surpassing human cognitive capabilities, AI is having its moment.

It’s also undeniable that AI will change how financial institutions and investors operate. However, the journey from “I want to use AI” to actually solving real business challenges using the technology isn’t straightforward (as of yet). In a recent Gartner report, the authors argue:

“Creating an AI strategy has become a distraction for many organizations and data and analytics (D&A) leaders. It is obscuring the focus that is needed on business outcomes.” (Gartner, 5 Practical Steps to Implement AI Techniques, 13 February 2023, Erick Brethenoux, Frances Karamouzis.)

The opportunities unlocked by AI are exciting, but careful planning will be required to capitalize on them. Organizations should start by selecting use cases that are connected to explicit business outcomes, and focus on delivering those outcomes rather than merely on deploying the AI tools themselves.

Gartner’s report proposes some approaches to use case selection. This blog post will provide an overview of these approaches as well as some concrete examples from financial services based on Alkymi’s experience helping our clients navigate the field of AI.

The basics of use case selection

Given the novelty and pace of innovation in AI, the most successful organizations will plan carefully before launching an AI strategy. To ensure the highest odds of success, your strategy should target use cases that are measurable, impactful, and feasible.

Measurable—can we track our progress?

The ability to establish objectives and track progress against them is essential to any business initiative. In the case of an AI strategy, this could take the form of a hard metric like “reduce overhead costs by 10% YoY” or a soft metric like “deploy intelligent document processing across at least two independent lines of business.”

Impactful—is this worth our time and resources?

The goal of any successful AI strategy is to deliver impact. Don’t launch a strategy that requires a disproportionate amount of effort to achieve its potential impact simply because it leverages AI

A good way to assess the impactfulness of your AI strategy may be to ask, “if we execute our strategy perfectly, how will it benefit the organization?” Unless it would be an overwhelming triumph relative to the strategy’s requisite resources, your strategy may need some tweaks.

Feasible—do we have a good chance of meeting our metrics?

If your AI strategy is sufficiently impactful and you have an adequate process in mind for measuring its progress, that doesn’t mean you necessarily have the ability to meet your objectives. For instance—if your strategy can’t succeed without increasing headcount in specific areas or building complex integrations, you may choose to prioritize other use cases.

With these foundations in mind, let’s look more closely at the process for specific use case selection in financial services.

Meeting the needs of your business

The broad potential of AI is not without immediate risks. The technology captures our imaginations and makes us wonder about all the profound ways it could transform our world, but also makes it difficult to identify use cases that are solvable given AI’s current stage of development. Given this challenge, it’s essential to partner with vendors who have expertise in your industry as well as in the development and use of AI tools themselves.

In financial services, some use cases are relatively feasible given currently available tools. Others are a stretch, and some are still nascent. Use cases also present distinct levels of potential impact.

In addition to these categories, more specific use cases powered by AI may be uniquely feasible (or may present unique impact) depending on what your organization can use them for.

For instance: a natural language answer tool that allows users to extract data from documents using simple, conversational commands can present tremendous value to organizations that handle large amounts of unstructured data. Specific tools like this are also relatively feasible to use with the assistance of a vendor with deep knowledge of your industry.

Some vendors even offer a suite of modular tools that can be structured to address your use cases. Organizations who partner with these vendors should first identify business needs which:

1. Are high-impact

2. Are able to be measured

3. Are not solvable using other tools

Equipping your team with clarity during the use case selection process is essential. The current state of AI means that it’s more important to choose your battles than to fight a war all at once. AI tools and vendors will evolve to address more use cases with time—but the time is now to start prioritizing areas where innovation can have impact.