SaaS Got Us Here, but Holistic Reasoning Will Take Us to the Next Level

Our industry's challenges—whether transaction decline, competitive density, or rising costs—are not new. What has changed, however, is the increasing reliance on technology to tackle these issues. The “SaaSification” of our industry has unlocked incredible capabilities and is promising operational transformation. Yet, paradoxically, it has also created inefficiencies by introducing silos. APIs offer some connection but often fail to align systems holistically across departments or goals.

The problem isn’t the systems themselves—it’s their inability to integrate with one another seamlessly. For example, a data record in one system might not mirror the fields of another system, adding complexity. While APIs (Application Programming Interfaces) are meant to solve these issues, they often require customization. Building or customizing APIs can create long engineering queues, with larger technology providers prioritizing their most prominent clients—usually larger ones. And since these larger clients tend to move slower, progress across the board grinds to a halt.

Even when APIs are implemented, they’re inherently limited. They allow one system to talk to another, but the communication is linear and siloed. It’s like being at a networking event where you can only speak to one person at a time—hardly an efficient way to collaborate.

Here’s a simple analogy: imagine you’re in a company's mailroom, responsible for tagging and forwarding thousands of letters to various departments. While the mailroom is essential, it’s disconnected from the phones, email servers, and chatbots where employees, customers, and vendors communicate daily. SaaS has made it easier to set up tools like mailrooms, call centers, CRMs, and chatbots, but these tools often operate in silos, leading to inefficiency and disjointed workflows.

So, how do we solve this?

The solution is to think of technology not as isolated systems but as a web—a connected ecosystem that seamlessly strings data together, regardless of the system. Data unification is the bedrock of progress. Some businesses achieve this by adopting a single, holistic platform that serves as a one-stop solution. But for many, this approach isn’t flexible enough to accommodate their unique needs. Microsoft CEO Satya Nadella recently highlighted this challenge, predicting that SaaS, as we know it, may soon evolve into something entirely different. He described SaaS applications as "CRUD databases with a business logic layer." What if we built models or agents instead of relying on systems? Models that learn and adapt as data enters your operations? By training these models, businesses could replace outdated processes and make smarter decisions that improve efficiency and effectiveness. Nadella suggested that this logic rapidly migrates to AI agents capable of reasoning, adapting, and orchestrating real-time operations.

These AI agents can move businesses from rigid systems to dynamic, interconnected ecosystems. By employing Chain of Thought (CoT) reasoning, AI can thread insights across multiple departments —connecting the dots to help organizations achieve unified objectives, circumventing individually biased goals, or departmental politics. When CoT reasoning connects workflows to objectives, AI transforms silos into networks of actionable insights. Businesses no longer operate as collections of disconnected systems—they become ecosystems that dynamically align every decision with overarching goals.

But there’s a catch, right?

At the heart of this shift lies data accessibility, cleaning, and tagging. These steps are foundational for building systems that can reason effectively and adapt to your business needs. However, challenges remain. Many existing systems house data points but make it difficult to access them, creating barriers to implementing a genuinely unified and intelligent workflow. Overcoming these challenges is critical to unlocking the full potential of AI-driven solutions.?

As Matt Fitzpatrick, our new CEO at Invisible, pointed out: “When great AI meets bad data, the data always wins.” This underscores the importance of clean, accessible data and high-quality AI training. Shockingly, only 8% of AI models make it into production, reflecting how crucial foundational data practices and well-trained models are in achieving meaningful outcomes. This training is accomplished through ‘Model Evaluations’ (Evals for short).?

To benefit from clear, demonstrable bottom-line impacts of Generative AI investments, organizations must commit to a data strategy that involves (a) committing to model development vs. continued investment into SaaS off-the-shelf solutions, and, more pertinently, (b) recognizing the need to improve data quality and the expected outputs generated from intelligent orchestration. No model out there will perform how you expect it without attention on these three core factors:

• Clean, Tagged Data: Businesses must invest in systems that ensure data is accurate, tagged, and accessible across all departments.
• Model Training and Evaluation: Models must be trained on diverse and relevant datasets, with rigorous evaluation to ensure they reason accurately across interconnected workflows.
• Continuous Learning: AI agents should evolve with changing inputs, retraining on new data to adapt to market trends and internal shifts.

Rearchitecting for an AI-Powered Future

Amazon’s transition into an AI-driven powerhouse offers a blueprint for organizations seeking to adopt bold changes in operating models. Before Jeff Bezos’s infamous memo above and their transformation, Amazon faced the same challenges as many firms: silos in its organization, data, and technology. These barriers stifled scalability and fragmented customer insights.

Bezos recognized the need for a radical rearchitecting of technology and organization. His vision was to build a centralized, software-driven platform to integrate operations, aggregate data, and enable advanced AI applications. This transformation wasn’t simple. Early iterations of the platform didn’t meet expectations, prompting Amazon to hire Brian Valentine, a seasoned software platform leader from Microsoft, to lead the redesign.

The result was the Santana platform—a modular, standardized architecture supported by agile, “two-pizza” teams (small enough to be fed by two pizzas). Santana preserved standard foundations, broke down silos, and ensured data was accessible across applications. This platform fueled innovations such as Amazon’s recommendation engine and Alexa, allowing the company to scale its AI capabilities rapidly.

Amazon’s transformation didn’t stop internally. With the rise of Amazon Web Services (AWS), the company democratized AI tools, enabling businesses of all sizes to leverage cutting-edge machine learning systems like SageMaker. This broader shift redefined an operating model, inspiring a generation of AI-driven firms like Ant Financial.

A Web of Intelligent Orchestration

True transformation lies in the interdependability of systems and decisions. AI-driven agents employing CoT reasoning help bridge departmental silos, enabling businesses to reason across data sets, goals, and workflows. This interconnected intelligence allows businesses to:

• Align marketing campaigns with operational capabilities, ensuring seamless execution.
• Use customer feedback to inform product design, staffing, and distribution decisions.
• Predict and adapt to demand fluctuations, reducing inefficiencies and costs.

When CoT reasoning connects these workflows, AI transforms silos into networks of actionable insights. Businesses no longer operate as collections of disconnected systems—they become ecosystems that dynamically align every decision with overarching goals. Let’s explore how that could support restaurants, retail, and CPG industries. Collaborative Demand Forecasting: No longer will sales data be analyzed in isolation. It will be integrated with feedback from marketing campaigns, distributor performance, and operational bottlenecks, providing a holistic forecast to optimize production schedules and minimize waste. This will flex levels of social media spend by geographies during a product release, influence labor schedules for cooks in kitchens, and provide real-time updates to suppliers to ensure stock-outs are a thing of the past.?

Localized Merchandising: Individual sales patterns based on time of year will identify buying personas and patterns that produce individualized planograms by location that update not on fixed quarterly schedules but based on the forecasted margin uplift anticipated against the cost of labor to undertake resets. Schedules and task planners will be automatically cascaded to local store GMs and prioritized against other routine tasks.? Upon completion, identified through camera scans of the newly merchandised shelves will trigger the initiation of geo-targeted marketing campaigns.?

Personalized Customer Experience: A frequent business traveler books a room. Customers checking in with their loyalty app at the start of an experience will only see items relevant to them. Marketing uses their preferences for quiet floors and particular amenities to offer curated packages. Vegans won’t see meat products on the digital tablet menu presented later, and the Spa will ensure allergies are already taken into account with the massage oil used. CoT reasoning pushes this data to housekeeping, operations, the spa team, and the restaurant, ensuring the room is prepared accordingly and guiding front-desk staff to offer tailored recommendations during check-in.

The Path Forward

The future of business technology isn’t about patching together disconnected systems; it’s about reimagining the entire framework. Agentic workflows offer a path forward, enabling systems that don’t just respond but proactively adapt, learn, and collaborate. This shift moves us beyond SaaS into an era of model-driven innovation where AI works seamlessly across every business touchpoint. Assuming SaaS platforms will eventually talk to each other is a long wait that may never transpire. Even holistic platforms that promise every feature under the sun have often acquired technology that doesn’t work well with the current infrastructure and, therefore, isn’t built to work seamlessly together. But neither SaaS or AI enabled capability will work optimally without data hygiene and appropriate quality assurance. This becomes more prescient if leaders are to empower holistic intelligent orchestration and the decisions it proposes within an operating environment.?

In this new paradigm, technology becomes less about tools and more about partners—designed not to add complexity but to remove it, and to address data disconnects. Leaders will need assurance that Generative AI capabilities not just work as desired, but that they derisk their operation from the inertia and captivity that their data ocean serves them today.? Businesses who embrace holistic chain of thought reasoning and the data standards required to stand it up can finally focus on delivering value, fostering creativity, and serving customers with unparalleled efficiency. The question isn’t whether we’ll embrace this future, but how quickly we can make it our reality.