In last week’s article, “AI Tokens, Model Selection and Cost Optimization: Building Smarter AI Systems“, we explored how organizations can optimize their use of Artificial Intelligence through better model selection, intelligent routing, and efficient token management. One of the key conclusions was that the effectiveness of AI is not determined solely by the sophistication of the model being used. The quality of the output depends heavily on the quality, relevance, and accessibility of the information being provided to it.
This naturally raises a broader question:
If AI is only as effective as the information it can access, are organizations focusing on the right problem?
Across the financial services industry, considerable attention is currently being given to selecting AI platforms. Firms are evaluating ChatGPT, Microsoft Copilot, Gemini, Claude, and the growing number of AI assistants being embedded within enterprise applications. Discussions frequently focus on model performance, security, cost, and productivity gains.
These are important discussions, but they do not fully address the deeper issue.
However, having spent many years helping organizations integrate systems, automate workflows, and establish connected reporting environments, Lee increasingly believes that the greatest challenge facing enterprise AI is not the model itself.
The deeper issue is that information often remains fragmented across the organization.
AI Is Not a New Conversation at Lima Capital
While the recent surge in interest surrounding Artificial Intelligence may make it seem as though AI has emerged overnight, the reality is quite different.
At Lima Capital, conversations around machine learning, language models, predictive analytics, and data-driven decision-making began many years before AI became a mainstream business topic.
Our early initiatives were not aimed at building general-purpose AI assistants; they were focused on investment intelligence.
As wealth advisors and investment managers working across fund structures, one of the challenges we identified early on was that investment decisions are rarely influenced by a single source of information. Market prices, economic indicators, company fundamentals, news events, sector trends, geopolitical developments, analyst research, and investor sentiment can all influence outcomes.
Even before the recent wave of AI adoption, we recognized that the future of investment analysis would increasingly depend on an organization’s ability to gather, connect, and interpret information from multiple sources.
This led to some of our earliest investments in intelligent data processing initiatives as far back as 2018.
Although the technology available at the time was far less advanced than the models we have today, the underlying challenge was already clear: the real value did not lie in collecting more data, but in creating context.
What made those early initiatives particularly significant was that they revealed a challenge that extended far beyond investment management.
Whether evaluating an investment opportunity, managing client relationships, monitoring compliance obligations, or overseeing business operations, meaningful decisions are rarely based on a single dataset. Instead, they are shaped by multiple sources of information that need to be connected, interpreted, and understood together.
Looking back, many of Lima Capital’s investments in automation, integration, reporting, workflow orchestration, and data architecture were not simply technology projects. They were foundational steps toward building environments where information could be connected and transformed into actionable intelligence.
In many respects, the recent AI revolution has not changed our direction; it has validated it.
The Financial Services Data Challenge
Having spent much of his career designing integrations, automating processes, and connecting business systems across financial services organizations, Lee has often found that the biggest challenges are not technological.
More often, they revolve around information existing in the wrong place, at the wrong time, or within the wrong system.
Whether working with investment managers, fund administrators, fiduciary businesses, or corporate service providers, the pattern is remarkably similar. Organizations invest in excellent technology platforms, yet the information within those platforms remains fragmented.
A typical wealth management or fund administration business may operate portfolio management platforms, CRM systems, accounting software, compliance and AML systems, document repositories, reporting tools, workflow applications, email environments, and collaboration platforms.
Each system performs an important role and contains valuable information, yet none of them tells the complete story.
Historically, this was manageable because people effectively acted as the integration layer.
Relationship managers gathered information before client meetings, operations teams consolidated reports, compliance officers reviewed data across multiple platforms, and management relied on reports assembled from information sourced throughout the organization.
The process was often inefficient, but it worked because people are naturally able to build context from fragmented information.
Artificial Intelligence changes this dynamic.
Intelligent Applications vs Intelligent Organizations
The current generation of AI tools is largely embedded within individual systems.
Microsoft Copilot understands Microsoft 365, CRM assistants understand client records, portfolio management platforms understand portfolios, and accounting systems understand financial transactions.
Each assistant adds value within its own environment, but none of them truly understands the organization as a whole. An AI assistant reviewing client correspondence may have no visibility into client profitability, while one analysing portfolio activity may be unaware of operational challenges or compliance concerns.
Similarly, an AI assistant reviewing financial information may have little visibility into client sentiment, service levels, or relationship risks.
In many respects, organizations are building intelligent applications while continuing to operate fragmented enterprises.
Whether that will be enough remains an open question.
One lesson that became clear during those early investment intelligence initiatives was that the quality of insights was rarely constrained by the analytical model itself.
More often, the real limitation was access to information.
A model may be capable of identifying patterns, correlations, and opportunities, but its effectiveness will always depend on the breadth and quality of the information available to it.
The same principle applies to enterprise AI today.
Even the most sophisticated model can only reason with the information available to it. If client data sits in one system, operational information in another, compliance records in a third, and reporting data somewhere else entirely, then the AI’s understanding of the organization will inevitably remain incomplete.
This continues to shape Lima Capital’s technology strategy today.
Building for the Next Generation of Enterprise Intelligence
At Lima Capital, our interest in Artificial Intelligence did not begin with AI models. It began with integration.
Over the years, we have seen organizations invest heavily in technology platforms designed to address individual business challenges. Portfolio management systems, CRM platforms, accounting solutions, compliance tools, document repositories, workflow applications, and reporting environments all play an important role within the enterprise, yet despite these investments, information often remains fragmented.
As technology teams, we frequently spend significant effort integrating systems, synchronizing data, automating workflows, and creating consolidated reporting environments. Historically, these initiatives were driven by operational efficiency, governance requirements, and management reporting needs, but today they are becoming equally important for AI.
The more we explore enterprise AI, the more apparent it becomes that the quality of AI-generated insights is directly linked to the strength of an organization’s integration strategy.
An AI model can only reason with the information available to it. If information remains isolated within individual systems, AI will remain limited to providing system-level intelligence. If information is connected, governed, and accessible, AI can begin delivering organizational intelligence.
This is one of the reasons Lima Capital continues to invest heavily in integration initiatives. Every system connected, every workflow automated, and every dataset made accessible adds context that can ultimately be leveraged by both people and AI.
In many respects, we view integration as a prerequisite for enterprise intelligence.
Beyond Data Access
When discussing enterprise-wide AI, many organizations immediately focus on permissions and security. These considerations remain critical, particularly within regulated industries such as wealth management and fund administration, but the conversation may need to evolve beyond simple data access.
Historically, permissions determined what information an employee could see. AI introduces a different possibility. A relationship manager may ask which clients should be engaged this week, and to provide the best answer, the AI may consider portfolio activity, service issues, onboarding delays, communication history, compliance reviews, client behaviour, and other operational factors.
The relationship manager does not necessarily need visibility into all of the underlying information, nor should they. However, the AI can still draw on that broader organizational context to provide a more informed recommendation.
In this model, permissions are no longer solely about what information a user can see. They also shape how AI interprets information and delivers guidance. The employee receives recommendations relevant to their role, informed by the wider organization, without exposing information they are not authorized to access.
The Emergence of AI Roles
This concept may ultimately lead organizations toward defining AI roles alongside traditional user roles.
A relationship management AI may focus on client retention, service quality, and growth opportunities, while an operations AI may concentrate on workflows, bottlenecks, and efficiency. A compliance AI, by contrast, may prioritize regulatory obligations and risk management.
Executive and board-level AI assistants may operate from a broader perspective, focusing on profitability, strategic objectives, organizational performance, and long-term business outcomes.
The same underlying data can therefore produce very different answers depending on who is asking the question, the role they perform, and the decisions they are responsible for making.
Perhaps the future is not a single AI assistant designed to serve everyone equally.
Instead, the future may lie in an organizational intelligence platform that understands the structure of the business, the objectives of each department, the responsibilities of each role, and the permissions associated with each employee.
From System Integration to Intelligence Integration
Historically, integration projects focused on moving data from one system to another, often through data warehouses, reporting layers, and business intelligence platforms designed to consolidate information for analysis. Today, that objective is evolving. As AI becomes increasingly embedded within organizations, integration is no longer just about data movement or reporting; it is about creating the connected context that enterprise AI requires.
At Lima Capital, we increasingly view this as the transition from system integration to intelligence integration.
The goal is not simply to ensure that systems communicate, but to create an environment where information can be connected, understood, governed, and ultimately transformed into actionable insight.
This does not require replacing existing systems. In fact, many organizations already possess the data they need; what is often missing is the ability to connect that information in a meaningful way.
Looking Ahead
As organizations continue evaluating AI strategies, the discussion will inevitably include models, platforms, security, costs, and productivity gains. These are all important considerations.
However, based on his experience integrating systems across financial services organizations and Lima Capital’s own technology journey, Lee believes the more important question may be whether an organization’s information ecosystem is capable of supporting enterprise intelligence in the first place.
The first generation of AI adoption has largely focused on helping individuals work more efficiently. The next generation may focus on helping organizations understand themselves more effectively.
At Lima Capital, we do not view Artificial Intelligence as a replacement for human expertise. Rather, we view it as a tool that helps our teams process information more effectively, identify risks and opportunities earlier, and make better-informed decisions on behalf of our clients.
The future belongs to organizations that successfully combine governance, integration, and intelligence. While much attention is currently focused on selecting AI models, the more strategic challenge may be creating the connected information ecosystems that allow those models to operate effectively.
The next generation of competitive advantage may not come from access to better AI. It may come from access to better-connected information, because ultimately AI is not replacing the need for integration; it is making integration more important than ever.
About the Author
Lee Zeinzinger is the Chief Technology Officer of Lima Capital and is responsible for the firm’s technology strategy, systems integration, automation, data architecture, and Artificial Intelligence initiatives.
With more than twenty years of experience across financial services, investment management, fund administration, fiduciary services, and enterprise technology, Lee has led numerous projects focused on integrating complex business systems, improving operational efficiency, and enabling data-driven decision-making.
Lima Capital views Artificial Intelligence as a tool to support human expertise, helping teams process information more effectively while preserving the judgment, experience, and accountability that remain essential to investment management.
By combining diverse expertise across investments, financial markets, operations, compliance, technology, and business strategy, Lima Capital continues to invest in integration, automation, analytics, and Artificial Intelligence to support better decision-making, improve efficiency, and ultimately help protect and grow client wealth over the long term.
