News
Comment
The evolution of language models is beginning to impact sectors that until recently seemed impervious due to their technical complexity. One such area is software development. The startup Sutro, founded in 2021 by Tomás Halgas and Owen Campbell-Moore, has introduced a platform that uses artificial intelligence to generate complete applications—for iOS, Android, and web—based on natural language descriptions, without requiring any programming knowledge. Sutro is leveraging artificial intelligence to enable the creation of full-fledged applications without coding.
Sutro’s architecture combines two approaches. On one hand, it uses GPT-4 and other language models to generate frontend and backend components. On the other, it incorporates proprietary rule-based compilation technology that translates high-level specifications into executable code. This hybridization aims to overcome a well-known limitation of purely generative models: their difficulty in maintaining structural coherence in complex systems requiring thousands of lines of code.
Traditional "no-code" platforms operate through drag-and-drop visual interfaces, yet still require users to possess detailed knowledge of the application’s structure. Sutro, by contrast, accepts conceptual instructions in natural language—such as "add a rating system to posts"—and automatically determines where and how to implement that functionality. Additionally, it allows for the insertion of custom code for users requiring finer control.
From our perspective in Information Technologies, this type of tool raises two key considerations. The first is standardization: by automating entire layers of development, generated applications are likely to share similar architectural patterns, which could facilitate maintenance but also reduce technical diversity. The second is governance: when a platform manages hosting, authentication, and security, responsibility for regulatory compliance (data protection, accessibility, etc.) becomes distributed between the user and the tool, demanding clear contractual delineation.
Automated software development will not replace traditional engineering in the short term, particularly for large-scale projects or those with highly specific requirements. However, for entrepreneurs, small businesses, and rapid prototyping, tools like Sutro represent a substantial reduction in entry costs and timelines. As with other generative AI applications, the ultimate value will not reside solely in the underlying technology, but in the platform’s ability to translate user intentions into functional, secure, and maintainable products.
However, I believe it will only be a matter of time before large teams of engineers are replaced by specialized AI-LLMs. This paradigmatic case may be the beginning of a major revolution that we will witness in the coming years.