The year 2024 was marked by immense possibilities in the field of artificial intelligence, but as we enter 2025, it's essential to turn those possibilities into tangible realities. For this to happen, cloud companies need to move beyond offering "primitive" tools, such as large language models (LLMs), and instead provide developers with practical, reliable solutions.

The current state of AI adoption is plagued by confusion and frustration. With new LLMs emerging every few days, each boasting improved performance, it's become overwhelming for mainstream developers to decipher which model to use and how to effectively integrate it into their applications. This has led to a situation where AI is often seen as a "WTH?!?" – a bewildering array of options that fail to deliver on their promises.

Simon Willison, a developer who has made significant strides in understanding and utilizing LLMs, has aptly described the challenges faced by developers. In his article "Things we learned about LLMs in 2024," Willison highlights the disparity between the marketed ease of use and the actual complexity of LLMs, which he likens to "chainsaws disguised as kitchen knives." He emphasizes that these models require a deep understanding and experience to wield effectively, making them inaccessible to many developers.

The lack of user-friendly AI applications is reflected in the biggest AI flops of 2024, as reported by MIT Technology Review. Even basic AI-powered features, such as Apple's text message summarization, have fallen short of expectations. This stems from the inability to trust AI to deliver consistent results and the burden placed on developers to turn AI primitives into functional applications.

The solution to this conundrum lies in the emergence of a company that can demystify AI and provide opinionated, easy-to-use platforms. Akin to Red Hat's role in simplifying open-source software or AWS's impact on cloud computing, the industry needs a player that can deliver trust, simplicity, and fewer moving parts. This would enable mainstream developers to harness the power of AI without getting bogged down in the intricacies of LLMs.

AWS and Microsoft are well-positioned to fill this gap. AWS's SageMaker, for instance, offers an integrated experience for analytics and AI, but its marketing description reveals a focus on providing a wide range of tools, which may overwhelm developers. Microsoft, on the other hand, has a history of providing intuitive, user-friendly solutions that cater to mainstream developers. By adopting a similar approach to AI, Microsoft could help drive widespread adoption.

The key to unlocking AI's potential lies in recognizing that the real money isn't in catering to the tech-savvy alpha geeks, but in providing accessible solutions for workaday developers. When cloud vendors shift their focus from AI primitives to user-friendly applications, AI will transition from a market rich in hope to one marked by widespread adoption and tangible results.