Forget the Tech—Your AI Strategy Starts with the Pain Point

We're seeing a concerning pattern with artificial intelligence adoption. Companies are starting with the technology first - testing different base models, evaluating every new AI tool, focusing on fine-tuning, and chasing the latest and greatest. But this approach fundamentally misses the mark, because all these technical decisions depend entirely on the problem you're trying to solve.

There seem to be countless fact-finding missions just to understand AI technology, followed by teams scrambling to figure out what to do with it. But this rarely works effectively. AI may be part of a particular solution - or it may not be needed at all. The essential starting point should be identifying where in your business automation opportunities truly exist.

The Parallel with Data Acquisition

At Nomad Data, we've seen this pattern before in data procurement. Most companies would start by examining various datasets, understanding their coverage and capabilities. Only much later would they ask: what do we actually do with this data? What's the outcome?

This backward approach wastes enormous time on evaluations that often aren't relevant or that solve "nice-to-have" problems rather than critical business needs. The solution isn't starting with the technology - it's starting with the problem.

This realization shaped how Nomad created its entire data discovery engine. We built it around use cases: start with a business problem that might need data to solve, then discover who has the appropriate data. Start with the problem, then discover the technology. In this case, the technology was the data itself.

We've applied this exact approach to artificial intelligence. We don't sell technology for technology's sake. Most end-users don't understand the technology, and frankly, they shouldn't need to. Instead, we interview business leaders to discover their pain points. Only then do we introduce our AI solution, Doc Chat, and sometimes augment it with data from our marketplace. But it always starts with the problem.

The Sales Process Transformation

In our early days, nearly every sales cycle began with the technology. Prospects would ask detailed technical questions: What can your Doc Chat engine do? How does it work with tables and graphs? How does it perform in specific scenarios?

But we quickly realized these conversations weren't productive because customers didn't have clear use cases. It was impossible to determine which features really mattered to them. Our breakthrough came when we flipped the sales process around.

We began by exploring people's pain points, understanding their business, and identifying friction points and inefficiencies. From that foundation, we offered solutions built on our Doc Chat technology that directly addressed their problems and were customized to their business. When you connect technology to a specific problem, you move much closer to delivering successful outcomes for clients.

Finding Automation Opportunities

Most of our solutions revolve around document processing. Businesses across industries interact with documents in countless ways, and humans spend significant time reading these materials. Insurance provides a perfect example.

The insurance process is overwhelmed with paperwork. During underwriting and application processing, companies receive numerous documents and must review them to understand the risk they're underwriting. This creates an enormous burden of document review and data extraction.

Similarly, on the claims side, adjusters often receive hundreds or thousands of pages that humans manually review to understand each situation. The fundamental pattern is consistent: a person examining documents - whether PDFs, Excel files, or other formats - to make decisions, gather information, or advance a process.

These activities typically don't add high value. They don't require sophisticated human thinking, making them ideal candidates for AI automation. The key identifier we look for is repeatability - processes that are well-defined enough that we can describe them to an AI system that could then perform them independently.

Overcoming Resistance to Change

When implementing business-first AI approaches, we encounter several types of resistance. Perhaps the most significant is that automation can appear threatening to certain roles - and sometimes, it genuinely is. If we're targeting a company with 50 people in document-intensive roles, and our solution can reduce task completion time from 10 hours to 10 minutes, those employees naturally fear for their jobs.

For some organizations, the goal isn't necessarily reducing headcount but rather preventing the need to hire additional staff as they scale. In other cases, the work being automated is so monotonous that it creates burnout and turnover problems. Automation can actually improve employee satisfaction by eliminating the most tedious aspects of their jobs.

We also encounter resistance stemming from previous negative experiences with AI tools. When you examine most roles closely, they're remarkably bespoke - unique to specific companies, industries, and even individual workflows. Generic, one-size-fits-all AI tools often struggle to recreate these specialized jobs because they're designed to perform many different functions but excel at none.

We address this challenge by building highly customizable engines that we can tailor to perform the exact tasks needed for each specific context.

When AI Isn't the Answer

AI excels at processing unstructured data, particularly text. While other AI types can work with images, sound, and other data forms, our focus remains on document-based artificial intelligence. This specialization means there are many scenarios where AI might not be the optimal solution.

When customers approach us with structured data needs - like extracting information from spreadsheets or databases - we often recommend traditional software solutions that perform better for those specific tasks. Large language models can technically handle structured data, but they're not necessarily the best tool for the job. The right technology depends entirely on the problem at hand.

The Discovery Process

Our approach to identifying automation opportunities begins at the organizational level. We first understand the various high-level functions within a business, then focus on areas that seem most amenable to AI automation.

The critical step comes next: interviewing the people performing specific jobs - underwriters, claims handlers, procurement managers, compliance professionals, legal teams - to thoroughly understand their workflows and where they spend their time. This quickly reveals whether AI is the right solution or if another technology might be more appropriate.

We maintain technological agnosticism throughout this process. While AI is undoubtedly powerful, many other powerful tools exist. What's particularly interesting is how AI has prompted a new wave of organizational introspection - companies are seriously reconsidering how they can improve efficiency across their operations.

Success Through Business-First Approach

One of our insurance clients provides an excellent example of successful business-first AI implementation. This company regularly faces litigation, receiving enormous volumes of documentation with each case - sometimes thousands or tens of thousands of pages. Previously, their staff had to meticulously review these documents, summarizing the litigation, understanding the claims, and tracking changes over time.

This process could take days or weeks and was extraordinarily tedious. Our solution enabled their staff to focus on understanding and analyzing the content by simply asking questions of an AI engine that could produce summaries and answer specific queries. The professionals could then make informed decisions about next steps, applying their industry knowledge and company expertise rather than wasting time on manual document review.

The results were dramatic: processing time collapsed from days or weeks to just 20-30 minutes per claim. This improvement not only saved time but also reduced risk by ensuring nothing important was missed in these documents - critical when litigation can involve claims worth $10 million or more.

Starting with Business, Not Technology

The key to successful AI implementation is starting with business stakeholders rather than technologists. Technologists naturally begin with technology; business people begin with business problems. Establishing a rough outline of the business problem before heavy technological involvement prevents wasted time researching irrelevant technologies.

The order of operations matters tremendously. It's similar to medical diagnosis - you wouldn't start with comprehensive body scans and blood work before discussing symptoms. You begin by understanding what's bothering the patient: What's your temperature? How do you feel? Then you determine the appropriate treatment approach.

When organizations successfully adopt this business-first mindset, implementation "just clicks." Once you understand the problem, communicating with technology teams becomes straightforward. You can clearly articulate whether a particular technology solves your specific problem. Developing proofs of concept, measuring their success, and justifying purchases or R&D investments all become significantly easier.

Practical Advice for Business Leaders

For business leaders feeling pressured to "do something with AI" but wanting to avoid the technology-first trap, our advice is simple: find a real use case as early in the process as possible. This makes everything more tangible and builds momentum and excitement.

The alternative - starting with technology and generating a list of 20 potential applications - typically results in doing all 20 poorly. Projects fail to reach even proof-of-concept stage, and enthusiasm quickly evaporates.

By focusing on solving genuine business problems first and considering technology second, organizations can ensure their AI investments deliver meaningful value rather than becoming expensive experiments. The most successful implementations are those where the technology becomes almost invisible, seamlessly supporting the business objectives that truly matter.