According to Gartner, ninety percent of CIOs say AI’s prohibitive costs are a limiting factor in AI … [+]
peopleimages.com – stock.adobe.c
According to Gartner, ninety percent of CIOs say AI’s prohibitive costs are a limiting factor in AI success. Another report, this one from KPMG, found that 68% of CEOs identify AI as an investment priority.
In other words, leaders are in a bind. They’re under pressure to implement AI, quickly prove the technology’s value and avoid the governance and security risks, all while figuring out how to shoulder the prohibitive cost for deploying AI across the organization. For many, it feels like an impossible math equation that just won’t work.
But what if there was another way to get started with AI that lowers the investment costs, demonstrates ROI and avoids the security risks? Executives need to go back to the basics, exploring how AI can solve specific business problems. AI agents—which can range from simple chatbots to process agents orchestrating and improving a process to complex adaptive systems working autonomously toward a goal—offer a promising road forward.
Read on for a practical AI approach that allows leaders to capitalize on AI’s groundbreaking benefits and test out AI agents securely—all without breaking their budget or embarking on a complicated year-long transformation project.
A Straightforward Approach to Practical AI Implementation
The shift to an AI-powered world has many leaders feeling overwhelmed. The technology is too powerful, moving too quickly and is too important to the organization’s future success. All those factors make it vitally important to get AI implementation right. That’s a lot of pressure.
Leaders need a small perspective shift. True, many companies are prioritizing AI investment, and some have successfully deployed the technology enterprise wide. But most are in the same boat: Still experimenting and figuring out how the technology best fits within the organization. Leaders may feel behind, but it’s a solid bet that most everyone else does, too.
Going all in on one AI tool or deploying it across the entire enterprise isn’t necessary or recommended at this stage. The important thing is to get started in whatever small ways make sense for the organization, its budget and goals. Here’s how:
Start with use cases. Before exploring specific AI tools, identify a few practical AI use cases. What are some problems AI can help solve right now? One way to go about this is to identify bottlenecks in the organization. Here are a few common ones:
- The valuable yet overextended team expert who has deep, niche knowledge. Requests for information can easily take these experts a month or two to address.
- Routine admin tasks that eat up too much time and cause project slowdowns.
- Software developers with too many requests for tedious tasks such as helping write test cases or document code.
- IT teams that resist creating and maintaining endless data reports for business performance managers.
- Senior team members responsible for passing information along to more junior employees, such as senior mechanics communicating how to fix an issue, or IT leaders answering questions about IT systems.
All these scenarios would likely benefit from an AI assist, such as a chatbot designed to speed up the transfer of information. Not only would this empower team members to quickly get the information they need, but it would also give overextended experts more time to focus on their most valuable work.
Run limited-scope pilots. To prove AI value, choose a use case and run a small, limited-scope pilot. This tactic offers a few benefits. The pilot’s small scale means lower costs. Limiting the pilot to one or two use cases makes it easy to measure and track ROI, and it can be done in a matter of weeks, leaving businesses with a quick, flexible experiment in how AI can benefit their organization. Rinse and repeat, building up internal expertise in AI along the way.
One of my company’s clients recently had success with a limited-scope AI pilot. An energy and utility company was bogged down with a costly, labor-intensive and slow process to inspect and maintain thousands of critical valves. These valves come in a variety of configurations from multiple manufacturers. Quality control technicians had to comb through hard-copy or PDF manuals to find the appropriate specs for each valve.
For the pilot project, the energy and utilities company focused on one manufacturer. My company deployed a proprietary AI agent to extract the appropriate information from the documents and build a user-friendly database. This allows quality control technicians to access specs much more quickly. The company picked a small area for AI experimentation, and now that the value is apparent, they are well positioned to repeat the process with other manufacturers.
Limit risks by creating a walled garden test atmosphere. Leaders can design narrow-scope AI pilots to operate completely within a walled garden environment. Designed this way, only a small number of people will have access to the data and the AI tool—ideally those who already have access to the data, such as the CFO and other top executives.
This setup limits the security and governance risks, providing a more secure way of testing the technology. It also gives leaders an opportunity to hone AI skills they can pass on to their direct reports once AI is more widely distributed throughout the company.
Creating a walled garden atmosphere does come with its own challenges. For instance, running AI within a closed environment means the AI tool can work only with a small set of data, rather than benefiting from data from the entire internet.
However, there are workarounds. Engineers from my company, Centric Consulting, have created accelerators (pre-built code or software tools) to solve this challenge in different toolsets. Whether an organization deploys AI internally or works with a third-party vendor, it’s important to first explore how a walled garden atmosphere might impact the use case, performance and cost.
It’s OK To Start Small With Practical AI
AI will undoubtedly change the business world in unimaginable ways. It’s a big technology with big promises. But that doesn’t mean companies have to figure it all out now. By starting with a small, narrow pilot project to determine clear ROI with AI or AI agents, leaders can get practical AI experience and prove value quickly—without a hefty budget.
