The video highlights the emergence of AI “wrappers” as companies shift their focus from foundational models to innovative user applications, exemplified by the startup Perplexity’s significant valuation increase. CEOs from Perplexity and Abridge emphasize that developing proprietary technologies is becoming more valuable, as decreasing costs in AI development enable startups to create and refine their own models, leading to a more competitive landscape.
The video discusses the evolving landscape of artificial intelligence (AI), particularly focusing on the emergence of AI “wrappers” and the shift towards user applications. It highlights the recent funding talks for the startup Perplexity, which aims to double its valuation to $18 billion. This development signifies a broader trend in AI, where the emphasis is moving from foundational models to the application layer, indicating that companies are now focusing on creating innovative user-facing applications.
Deirdre Bosa, the presenter, notes that the application layer has officially arrived, and model building is becoming more cost-effective and efficient. This combination is challenging the notion of AI wrappers, which were previously seen as mere interfaces over foundational models. The video emphasizes that the true innovation is happening at the application level, as demonstrated by Perplexity’s significant valuation increase in less than a year.
The CEO of Perplexity, Arvind Srinivas, argues that the perception of being just a wrapper will eventually fade. He asserts that his team is capable of building and serving their own models, which reflects a deeper level of innovation than previously recognized. This sentiment is echoed by Shiv Rao, CEO of another AI company called Abridge, which focuses on simplifying medical paperwork. Rao believes that the term “wrapper” is outdated, as companies are now realizing that value can be derived from developing proprietary technologies rather than solely relying on large-scale foundational models.
Both Perplexity and Abridge are actively building and fine-tuning their own AI models, leveraging proprietary datasets to enhance their offerings. The video points out that the decreasing costs associated with AI development are facilitating this shift towards the inference stage, allowing more startups to emerge in the application layer. This trend could lead to increased funding opportunities for AI app startups, as evidenced by Perplexity’s current funding discussions.
In conclusion, the video illustrates a significant transformation in the AI industry, where the focus is shifting from foundational models to innovative applications. As companies like Perplexity and Abridge develop their own technologies and refine their models, the landscape of AI is becoming more competitive and diverse. This evolution suggests a promising future for AI startups that prioritize application development and leverage the decreasing costs of model building.