Dottxt Raises $11.9 Million to Teach AI Models Structured Communication

As AI continues making strides in conversational abilities, the way it interacts with software engineers and delivers information remains an obstacle for widespread adoption. Dottxt, a startup developing tools to enable structured communication between AI systems and developers, has raised $11.9 million to advance its solutions.

The company was founded by the creators of Outlines, an open-source project allowing more nuanced guidance of large language models without dependence on prompt formatting. By describing desired outputs instead of crafting specific questions, it gives AI a framework for responding logically.

However, Outlines only supports Python while Dottxt aims for model-agnostic cross-platform use. It also adds features like performance optimization absent from the original framework. With 2.5 million downloads showing developer enthusiasm, Dottxt's founders saw an opportunity to commercialize their .

Their approach leverages the idea that most digital content follows patterns like code syntax or document templates. By describing a “grammar” for AI to follow, it can answer as intended for applications. Microsoft, OpenAI and others now recognize the potential of this “structured generation” method.

Raising $11.9 million in a seed round led by EQT Ventures enables Dottxt to build out their platform and hire more staff. Currently testing with enterprises, the goal is accelerating adoption within 18 months rather than immediate monetization.

As AI is increasingly incorporated into software, the ability to integrate models as seamlessly as other tools is important for widespread use. Dottxt aims to “make AI speak computer” through an interface addressing a key challenge holding back broader AI deployment. With developer and investor interest in structured communication growing, their approach could become integral to advancing generative technologies.