The Principles for Achieving Lift with Sovereign AI

AI success, much like powered flight, follows a set of fundamental principles. The Wright Brothers, through relentless experimentation, discovered that lift (L) must equal weight (W) to maintain altitude, while thrust (T) must exceed drag (D) to propel an aircraft forward. This formula is a given nowadays, but it took 50 years—from the first glider flight in 1853 to the first powered flight in 1903—to refine it.

Even with the advent of the combustion engine in 1885, it took another 18 years to integrate thrust into the existing principles of gliding flight. Yet, in 1919, just 16 years after the Wright Brothers’ breakthrough, the first transatlantic flight was completed. This acceleration was remarkable—a threefold faster evolution than the initial learning cycle. Hundreds of aircraft companies emerged within that short window, and some of them, including Boeing (NYSE:BA) and Curtiss, still exist today. Some of these early forms of flight, such as the Brazilian Santos-Dumont 14-bis biplane, were at best flimsy and unscalable from an engineering architectural perspective.

Similarly, early AI implementations may be rudimentary, but those with scalable architectures will find long-term success. AI, like aviation, will evolve rapidly, compressing decades of progress into just a few years. Getting the fundamental principles in place for flight should be your focus.

The first principle: Learn

The capacity to constantly learn and adjust in real time is what drove success or marked failure for most of these early powered-aircraft designs and companies. Those that survived continuously learned and iterated—incorporating new technologies, responding to real-world conditions, and evolving based on data.

AI follows the same trajectory but at an exponentially faster rate, just the way the aircraft industry grew, with Whittle’s jet propulsion engine of 1930 seeing its first real use in 1941.

“AI teams cannot afford to lose years, even decades, in experimentation,” says Nancy Hensley, Chief Product Officer at EDB, a leader in enterprise Postgres data and AI solutions. The pace of data generation is staggering—it doubles every few months. The key challenge is not the volume of accessible data itself but creating the ability to observe, integrate, and act on it effectively.

“The future of AI belongs to organizations that become learning engines. Success demands observability across data estates, secure sharing of insights, and a culture of continuous adaptability. Companies that build AI not just as a tool but as a strategic and sovereign platform will lead the next phase of AI-driven business growth,” says Hensley.

Take a simple example. A bank has a new customer-facing AI agent designed to radically reduce Level 1 support. However, the same AI insights can later optimize fraud detection, enhance cybersecurity, improve marketing timing, and even mitigate customer churn.

That ability to transfer insights across functions is a vital part of the constant learning imperative. You can see this pattern in successful organizations facing digital transformation too, especially in financial services and banking, where tight regulatory compliance requirements demand a particular focus on the need to learn, share, and experiment in a focused manner.

Platform design principles: Observability, governance, and scale

According to research from IDC, more than 90% of enterprise data is unstructured and siloed within databases that lack observability across the data estate. Without a layer of access and integration, organizations struggle to leverage AI effectively.

Hensley suggests that open-source technologies such as PostgreSQL, when deployed at an enterprise scale, offer the extensibility and flexibility to scale with enterprise observability and governance requirements.

In a 2024 survey of hundreds of data leaders across the U.S. and U.K., EDB found that 67% of enterprises now operate in hybrid environments and integrate AI capabilities across their Postgres estate.

“To achieve scalable AI success, organizations must embrace a data and AI platform mindset. This is not just a technology decision—it represents a fundamental shift in how data is accessed, structured, and leveraged,” said Hensley.

The engine for AI success

Just as early aviation engineers debated monoplane versus biplane designs, AI leaders must decide how best to architect their data and AI platforms for scale and adaptability.

“Beyond a strong data foundation, organizations need AI engines and applications that continuously learn, apply insights at rapid speed, and evolve as a core principle of business success,” says Hensley.

According to Hensley, tools such as Griptape AI (part of the EDB Postgres^® AI platform) allow companies to protect their sovereign data, rapidly infuse data and generative AI models across Postgres estates, and build sovereign platforms with minimal technical barriers or code construction.

Consider how generative AI can work across a financial institution:

• Enhancing fraud detection and cybersecurity by analyzing real-time transactions

• Optimizing marketing campaigns by identifying behavioral patterns in customer interactions

• Improving customer service through intelligent AI agents that continuously learn from interactions

The fundamental principle of AI success is simple: Organizations must become learning engines powered by a strong data and AI platform, driven by continuous adaptation. Then, the physics of AI success will take over, just as with powered flight, accelerating progress at an unprecedented rate. Those who build the right AI and data foundation today will lead in the era of intelligent automation tomorrow.