The enterprise AI strategy that outlasts any single model

In January of this year, few enterprise tech leaders would have bet on Anthropic over OpenAI. Today, Claude reigns supreme (inspiring a notable 180 by Elon Musk), with Gemini threatening to take market share and introduce pricing models that could flip the leaderboard on its head again. That’s exactly why betting on a single model is a dangerous strategy.

The most successful organizations won’t be those trying to guess tomorrow’s top-tier model, nor will they wait passively for future releases. Instead, they will invest in underlying frameworks that continuously improve regardless of which specific AI model drives them.

Why betting on one AI model is a losing strategy

Our strategy for AI, through recursive self-improvement (RSI), is rooted in this core principle. RSI is an approach to AI that compounds its own abilities by improving itself. If done carefully, RSI can function as an overarching layer above any model. Crucially, given RSI’s inherently compounding trajectory, it represents the most likely contender to be the approach that reaches superintelligence, no matter which model is used underneath.

Though recently achieving the status of a Silicon Valley buzzword, applying something like RSI to unlock superintelligence has been the Holy Grail of AI research for decades. It’s what researchers like us have recognized since the 1960s as a critical step along the path towards what we call artificial superintelligence (ASI) today.

Recursive self-improvement compounds value beyond the model

The fundamental premise of RSI is that the next phase transition in AI won’t come from a system that has been taught to improve by any of the traditional methods of the past few years. Relying purely on data, compute and human intuition to generate exponentially improving capabilities is a path with hard physical and practical limits. Instead, the leap will come from a system that invents its own improvements and feeds them back into itself.

RSI has already delivered what business leaders would recognize as a virtuous cycle. Each improvement increases the system’s capacity to generate the next improvement. Competitive advantage compounds because the system benefits from both its own recursive progress and every improvement in the underlying models.

As Anthropic puts it, AI that can improve itself would be a “major development in the history of technology.”

Indeed, we believe it’s the single most important frontier of AI research. Anthropic’s model-specific approach to RSI is already paying off for them: a recent Anthropic Institute report captures the pace of change with real world impact, “Claude-written code was somewhat worse than human-written code at Anthropic in late 2025, is roughly at parity today, and we expect it to be strictly better within the year.”

That’s worth applauding.

But what about the companies not currently building an in-house model? For them, looking at improvements that only take place inside the model someone else develops is unnecessarily limiting. It makes more sense to embrace a model-agnostic approach to RSI that improves whenever any new model is released.

Approaching the model as one component of a system without relying on any individual provider or tool makes it possible to achieve recursive improvement at the system level. By using an RSI approach that works outside the model and can swap models instantaneously, companies can immediately benefit from the compounding effects of self-improving AI.

Build a model-agnostic AI strategy that benefits from every breakthrough

This holistic approach to RSI fits the market today. The AI landscape is becoming more dynamic by the month. Frontier models leapfrog one another, open-weight models improve at remarkable speed, pricing strategies change and entirely new capabilities emerge in rapid succession. For enterprise leaders, the lesson isn’t to predict the next winner. It’s to build systems that improve regardless of which model comes out ahead.

In fact, organizations that tie their future to a single model provider risk getting left behind altogether if a different model’s next iteration leapfrogs the one they’ve signed a long-term contract to use.

Every enterprise tech leader already understands the power of virtuous cycles. Amazon didn’t build an enduring competitive advantage by betting on a single product. Every improvement to its logistics network attracted more sellers, which increased selection and order volume, which justified further investment in logistics. Visa became more valuable as more merchants accepted its cards, attracting more cardholders, which, in turn, encouraged even more merchants to join.

Organizations that anchor their AI strategy to a single model provider will spend the next decade reacting every time the frontier shifts. On the other hand, organizations that build model-agnostic systems will benefit from every shift. Every improvement from Anthropic, OpenAI, Google, Meta or the next breakthrough model becomes another source of competitive advantage.

The world’s most durable companies don’t simply accumulate assets – they build systems where every improvement makes the next improvement easier; creating long-term advantage.

Enterprise AI should be approached the same way.

For enterprise tech leaders, that’s the strategic shift that matters. Stop asking which model deserves your long-term bet and instead, start asking whether your AI strategy creates its own virtuous cycle.

This article is published as part of the Foundry Expert Contributor Network.
Want to join?