Artificial intelligence has become almost impossible to avoid in the business software market. Over the past few years, AI has moved from an emerging technology discussed primarily by technology teams to a boardroom priority that is now influencing software purchasing decisions, corporate strategy, and even the questions organizations ask in an RFP. It is increasingly common to see software vendors promoting AI as a central part of their value proposition, while buyers are asking whether a product is AI-powered before they have necessarily established what they want the AI to accomplish.
There is a great deal of justification for this enthusiasm. Modern AI systems are remarkably capable. They can summarize large amounts of information, identify patterns, generate content, answer questions using natural language, assist with research, and automate tasks that previously required significant amounts of human effort. The technology is advancing rapidly, and it would be difficult to argue that AI will not fundamentally change how many businesses operate.
However, the rapid adoption of AI creates an important distinction that is sometimes lost in the excitement: the ability to do something with AI does not necessarily mean that doing it with AI creates business value.
That distinction has shaped one of the more deliberate decisions we have made at AdmiralBridge. While we are actively following the development of AI and exploring where it may eventually provide meaningful value to our customers, we have intentionally chosen to build the core AdmiralBridge product without AI features.
That decision is not a rejection of artificial intelligence. It is a decision about where we believe AI belongs, what our customers actually need, and what standards we believe enterprise software should meet before AI becomes part of a business-critical workflow. We also work for the Loss Prevention industry. An industry that prides itself on truth, accuracy, accountability, privacy, and data security. These things make the AI conversation a difficult one.
The gap between AI adoption and AI value
The business world is clearly moving toward AI. McKinsey’s 2025 State of AI research found that 88% of organizations surveyed were regularly using AI in at least one business function, a significant increase from the previous year. The same research, however, illustrates a much more complicated picture when the discussion moves from adoption to measurable business impact. Only 39% of respondents reported an enterprise-level EBIT impact from AI, and most organizations had not yet reached the point where they could say that AI was fundamentally transforming their enterprise.
The Canadian market provides an even more interesting perspective. KPMG Canada reported in 2025 that 93% of Canadian organizations surveyed were using AI in some form, compared with 61% the previous year. Yet only 2% reported seeing a return on their generative AI investments. That does not mean the technology is failing. It does, however, demonstrate that there can be a considerable difference between implementing AI and actually generating measurable economic value from it.
This distinction is important because software companies have a natural incentive to emphasize capability. A new AI feature is easy to demonstrate. It is easy to put into a product announcement, a sales presentation, or an Request for Proposal response. The harder question is whether that feature changes the economics or effectiveness of the customer’s business in a meaningful way.
A business does not benefit simply because its software can generate an AI summary. It benefits if that summary allows an employee to complete work more accurately or more quickly. It does not benefit simply because an AI model can identify a pattern. It benefits if identifying that pattern results in a decision that improves an outcome. The technology is only valuable when it translates into something the organization actually cares about.
That is the standard we believe enterprise software should be held to.
Trust becomes more important as the stakes increase
The question becomes even more complicated when AI is introduced into systems that manage information people rely on to make consequential decisions.
Consider the type of information managed by a loss prevention or incident management platform. A single case may contain employee statements, investigation notes, photographs, financial information, dates, locations, supporting documentation, witness information, and the history of actions taken throughout an investigation. That information may ultimately be used by investigators, loss prevention leaders, human resources teams, legal departments, compliance teams, or executives.
In that environment, the quality of the system cannot be measured simply by how quickly it produces an answer. The organization also needs to know whether the answer can be trusted.
That distinction is becoming increasingly important as businesses gain more experience with generative AI. A global study conducted by the University of Melbourne and KPMG, involving more than 48,000 people across 47 countries, found that while 66% of respondents regularly use AI, only 46% said they were willing to trust AI systems. The study also found that 56% reported making mistakes at work because of AI, while 66% said they rely on AI output without evaluating its accuracy.
Those findings should not be interpreted as an argument against AI. Instead, they demonstrate why organizations need to think carefully about the relationship between AI output and human decision-making. An AI system can be extremely useful while still requiring meaningful oversight. The appropriate balance depends on what the system is being asked to do and what happens when it gets something wrong.
The consequences of an inaccurate AI-generated marketing headline are very different from the consequences of an inaccurate interpretation of an incident investigation. In one case, a person may simply edit the text. In the other, an incorrect output could influence an investigation, an escalation, a compliance decision, or an employment-related action.
The higher the consequence of an incorrect answer, the higher the standard for introducing AI into that workflow.
Reliability is not simply an accuracy percentage
One of the challenges with discussing AI reliability is that accuracy can sound like a simple technical measurement. A system that is described as being 98% accurate sounds impressive, but that number alone does not tell a business whether the system is appropriate for a particular task.
The important question is not only how often an AI system is correct. It is also what happens when it is incorrect.
If an AI system is helping an employee draft a routine communication, the employee can review and correct the output. If the system is summarizing a meeting, the consequences of an error may be relatively limited. But if the same technology is being used to make a recommendation about an investigation or to identify a potential compliance issue, the organization needs considerably more information about how that output was produced and how it will be validated.
This is where trust, reliability, and business value intersect.
A business needs to understand whether the output is sufficiently reliable for the task. It needs to understand whether a human can identify and correct errors. It needs to understand what information the AI has access to. It needs to understand how the system handles uncertainty. It needs to understand what happens when the model produces an answer that sounds convincing but is wrong.
These are not theoretical questions. They are practical questions about risk.
As Deloitte’s research into enterprise AI adoption has highlighted, organizations that are getting meaningful value from AI are increasingly focused on redesigning workflows, establishing governance, and determining where human oversight is required.
The organizations creating value are not simply adding AI to existing processes. They are considering how the process itself should change.
That distinction is one we take seriously at AdmiralBridge.
Sometimes the right answer is not AI
Our core product exists to solve a relatively straightforward business problem: helping organizations manage incidents, investigations, cases, corrective actions, reporting, and the associated information in a structured and secure way.
Many of those problems do not require artificial intelligence.
A workflow can be automated through clearly defined rules. A report can calculate a result using deterministic logic. A permission model can determine who should have access to information. A case can be escalated because a defined condition has been met. A corrective action can be assigned because a workflow requires it.
There is considerable value in software that behaves predictably. In many enterprise applications, predictability is not a limitation. It is an advantage. When a customer creates a case and the system executes a defined workflow, the customer should be able to understand why that happened. When a report calculates a total, the customer should be able to trace the result back to the underlying data. When a security rule restricts access to information, the organization should understand what the rule does and why it exists.
Not every business problem benefits from introducing probability into the equation.
Enterprise-grade case management should be accessible to more than the largest retailers
There is another reason for our decision, and it is closely tied to why AdmiralBridge exists in the first place. We believe enterprise-grade case management should not be reserved for the largest retailers.
A retailer with several thousand locations may have an enormous loss prevention organization, dedicated technology teams, and substantial software budgets. A smaller retailer may have a much smaller team and significantly fewer resources, but the underlying business requirements are remarkably similar. Both organizations need to document incidents, manage investigations, protect sensitive information, track corrective actions, analyze trends, and provide leadership with reliable information.
The difference should not be whether one organization can afford to buy every new technology feature that appears in the enterprise software market.
Our goal has always been to make AdmiralBridge accessible to the small and medium-sized retail market while still providing the capabilities and security expected from enterprise software. Keeping the core platform focused and affordable is therefore not simply a pricing strategy. It is part of our product philosophy.
We do not believe every customer should automatically pay for AI infrastructure, AI functionality, or AI-related capabilities that may not provide value to their organization.
If an organization wants AI-enhanced functionality and believes it will generate a return, that should absolutely be an option. But the organization should be able to make that decision based on its own requirements rather than being forced into it because every software vendor decided that AI needed to be part of the base product.
Organizations deserve a choice
There is also a broader principle behind our approach: organizations should have a choice about whether they want to participate in AI-enhanced software.
Different companies will make different decisions, and there are legitimate reasons for doing so. Some organizations will want to adopt AI aggressively and integrate it throughout their operations. Others may have internal policies that limit the use of generative AI. Some may have concerns about privacy, intellectual property, data residency, regulatory requirements, or how information is processed by third-party models. Others may simply conclude that they do not currently have a sufficiently compelling business case.
All of those positions can be reasonable. A software vendor should not assume that its customers have made the same strategic decision about AI that it has. This becomes particularly important when software is managing sensitive business information. Customers should be able to clearly understand what happens to their data, what technologies are processing it, and what role those technologies play in the application.
For some organizations, the answer they want from a software vendor is not a description of the latest AI model or its capabilities. It is simply:
No, your data is not being sent to an AI model as part of this product.
There is value in that answer. Choice, transparency, and control are just as important as innovation.
We are not ignoring AI
It would be easy to interpret an AI-free product as a company that has chosen to ignore artificial intelligence.
That is not what we are doing. We are watching the technology closely, experimenting with it, and spending considerable time thinking about where it can provide genuine value to our customers. The fact that we have not added AI to the core product does not mean that we believe AI has no place in AdmiralBridge. It means that we believe the standard for introducing it should be higher than simply having the capability available.
If we introduce AI into a workflow, we want to understand the problem first. We want to know what the customer is trying to accomplish, what makes the current process difficult, and whether AI actually improves the outcome. We want to understand what happens when the system is wrong and what level of human oversight is appropriate. We want to know whether the customer wants the capability and whether the resulting improvement justifies whatever additional complexity and cost it introduces.
The next phase of the AI conversation
The first phase of enterprise AI was largely about possibility. Organizations wanted to understand what the technology could do, and software companies rushed to demonstrate those capabilities.
The second phase has been about adoption. Companies are deploying AI, employees are experimenting with it, and executives are trying to determine how it fits into their technology strategies.
The next phase needs to be about judgment.
Organizations need to become much more deliberate about deciding where AI belongs and where it does not. They need to understand the consequences of failure, establish appropriate governance, determine where human oversight is necessary, and measure whether the technology is actually creating value.
That will require a different kind of conversation.
- Instead of asking only, “Can AI do this?” we need to ask, “Should AI do this?”
- Instead of asking only, “How quickly can we deploy it?” we need to ask, “What business outcome are we trying to improve?”
- And instead of asking only, “How accurate is the model?” we need to ask, “What happens when it is wrong?”
Those questions are harder, but they are ultimately more important.
Technology should follow the business problem
At AdmiralBridge, our approach to AI is really an extension of how we approach software development as a whole. We start with the customer problem, understand the workflow, identify where friction exists, and then determine what technology can genuinely improve the process. That philosophy does not change simply because a new technology happens to be exciting or because the market has created an expectation that every modern software product should include it. We believe AI will become an important part of enterprise software, and there are undoubtedly problems where it will be extraordinarily useful and workflows that will be fundamentally improved by its application. We want AdmiralBridge to be part of that future, but we do not believe every problem requires AI. Good software should first be reliable, understandable, secure, accessible, and valuable to the people using it. When AI is eventually introduced into a business-critical workflow, we believe it should be there because it makes that workflow meaningfully better for the customer, not simply because the market expects another checkbox on an RFP.
So, what’s next?
We are comfortable with the decision we have made today, and the core AdmiralBridge platform will remain AI-free because we believe our customers deserve an affordable, reliable, and predictable case management solution, while also having the freedom to decide if and when AI becomes part of their own operations. That said, being deliberate does not mean standing still. Behind the scenes, we have been exploring a few ideas where we believe AI could solve a very specific problem and deliver meaningful value without compromising the reliability and control our customers expect from AdmiralBridge.
We are taking our time because when AI becomes a part of the platform, we want it to have a clear purpose and meet the same standards we apply to everything else we build. I don’t want to give too much away just yet, but we are working on something that could make some of the information already sitting inside AdmiralBridge considerably more useful to the people who need it. For now, I’ll leave it at that. We don’t believe software needs AI simply because AI is available. We believe AI needs a reason to exist, and we think we may have found one.
One thing will always be true, AdmiralBridge will always offer an affordable case management solution. however, if you are interested in exploring a bit deeper….. Stay tuned.
