In recent years, the artificial intelligence (AI) landscape has evolved from mere curiosity to incessant noise. Conference slogans, vendor solicitations, and slideshows all seem to start with the same question: What can AI do for you? And too often, the answer comes in the form of a catalog of hundreds of “use cases,” carefully presented, context-free, and ready to be plugged into any organization that accepts that transformation can start with a menu.
1898 & Co., part of Burns & McDonnell, takes the opposite view: AI is not a destination but a powerful tool to use to solve particular types of problems. The first question is not what the customer wants to order, but what problems they are looking to solve. It is from this that the right approach to the challenge and the appropriate toolkit for the job are developed.
Start with the problem, not the platform
This mindset is emblematic of how we approach customer needs and engineering, data and now AI. Technology should never be a destination. AI itself is not the deliverable. It is one tool among many that helps us achieve meaningful and measurable results. When applied correctly, AI can be transformative, while when applied indiscriminately, it may well represent another expensive experiment doomed to never reach production.
The work begins long before a model is selected or an algorithm is verified, developed, or tuned. It is essential to start by understanding the business challenge at hand. This means working directly with technical domain specialists in production, transmission, manufacturing, or any other environment where operational decisions are important. It is imperative to define the problem, the constraints, the desired results and the conditions under which a solution must work.
From there, the reality of the client’s data landscape and systems must be assessed: what information exists, where it is stored, and how it can be transformed, connected, or augmented. Gaps and obstacles must be identified to determine how to move forward.
That’s when it’s time to use the tech tool belt. Sometimes the optimal answer is AI. Other times it’s advanced analytics, automation, or machine learning. In most cases, it’s a combination, all orchestrated to solve a problem rather than to demonstrate a technology. Solutions must be designed to scale responsibly, improving operational reliability rather than compromising it. The pilot is not about “demonstrating” but about de-risking: solving the core problem in a controlled environment, creating clarity rather than hype.
This approach may seem simple, but it’s what differentiates successful AI programs from stalled ones. AI is a means to an end; it is not an end in itself.
Researching use cases
Last year, a client came to us with a familiar request: provide us with a list of AI use cases. Several large consulting firms had already presented compendiums of hundreds of possibilities, described in abstract terms and presented for maximum excitement. Of course, we also had such a list. However, as the dialogue continued, it became increasingly clear that a list was not getting us closer to the client’s goals.
After all, no customer needs hundreds of solutions. What they need are tangible, practical answers to real business and operational challenges.
Once we moved beyond the high-level solicitation and engaged in conversations with operators, asset managers, engineers, and data teams across the organization, it became clear that the real opportunities lay behind the day-to-day operational issues. No actor on the ground requested AI. They sought help to resolve problems so entrenched they were considered permanent. Inconsistent data, inaccessible documents, duplicate records, untraceable assets, and information that took hours or days to locate. All common problems with new solutions thanks to emerging and AI technologies. The path forward therefore became obvious. Even though AI was no longer the focus, for this set of challenges it proved to be the most effective tool in the toolbox for a range of data extraction, organization, and remediation challenges.
This problem-driven approach ultimately resulted in a highly focused pilot project that solved one of the organization’s biggest operational bottlenecks: generating clean, complete, and reliable data on their production fleet assets. And the proof of concept wasn’t just a throwaway technology demonstration, it solved the problem, delivering validated asset hierarchies far faster than the client thought possible. In just a few months, this pilot project grew into a $1.3 million implementation, accelerating the maturity of the client’s asset data environment and thereby improving the reliability of its operations and maintenance (O&M) generation activities. What they hoped would take years to accomplish – if it was indeed achievable – was accomplished in months.
As is often the case when real problems are solved, the project revealed new opportunities where AI could significantly reduce effort, mitigate risk, and finally tackle challenges deemed too costly, too complicated, or where data quality was deemed too poor to tackle. BOMs, attributes, and work order automation were all, for the first time, on the table for the customer and for us to apply AI tools to deliver. Success proliferated not because we pushed technology but because we followed value.
The 1898 & Co approach.
AI, for us, is never the starting line. It’s never the product. It is a mechanism for solving important problems: problems related to security, reliability, compliance, productivity and costs.
Our customers don’t need another slideshow full of possibilities. They need solutions grounded in business logic, designed for operational reality, validated by domain experts, and designed to scale responsibly.
We approach AI the same way we approach engineering: by defining the problem, understanding the system, selecting the right tools, and proving the value in controlled increments. The results speak for themselves, as AI becomes a capability rather than an experience; an asset, not a trend.
At 1898 & Co., we will continue to develop AI in this way: problem-first, results-driven, and domain-aligned. We don’t help customers apply AI, we solve problems with the new and advanced tools that increasingly populate our technology environment. Increasingly, we have the right tools to optimally solve an ever-widening range of problems.
—Chris Wiles is an AI Solutions Architect at 1898 & Co., part of Burns & McDonnell, specializing in applying AI to solve complex operational challenges in energy and infrastructure.