In a bold move to bridge the gap between generative AI and traditional software engineering, Dottxt has raised $11.9 million in funding to help AI models communicate more effectively with developers. The startup, founded by the team behind the open-source project Outlines, aims to eliminate the frustrations that come with using large language models (LLMs) by making them “speak computer.”
As enterprise CIOs cautiously adopt generative AI, one key challenge has been the technology’s struggle to integrate with existing workflows. Developers often find themselves using cumbersome tactics to coax valid outputs from LLMs, like resorting to overly dramatic prompts. Dottxt seeks to change that by enabling a more structured approach to AI interactions.
The startup’s flagship offering centers on a technique known as structured generation, allowing developers to guide AI outputs in a more sophisticated manner than simple prompt manipulation. “It’s about telling AI models how to answer,” says Dottxt CEO Rémi Louf. By focusing on the quality of AI outputs, Dottxt aims to restore a traditional engineering workflow where grammar and structure are refined for optimal results.
Dottxt’s approach is model-agnostic, promising better performance and a richer feature set than its predecessor, Outlines. Louf, a Frenchman with a PhD and a background in Bayesian statistics, believes their grounding in probability theory has illuminated the potential of structured generation. He points out that most of the text we interact with is structured—ranging from code to various templates—making it essential for LLMs to follow these formats to be truly useful in professional settings.
The demand for structured generation has already caught the attention of major players, including OpenAI, which recently introduced its own version of structured outputs, inspired by Outlines.
Investors are taking note of Dottxt’s potential. The startup secured a $3.2 million pre-seed round led by deep tech VC firm Elaia in 2023, followed by an impressive $8.7 million seed round led by EQT Ventures in August. This funding surge comes on the heels of a remarkable increase in the popularity of Outlines, which has been downloaded over 2.5 million times.
With plans to expand their remote team from eight to 17 by the end of the month, Dottxt is prioritizing developer relations and ecosystem building. Louf envisions the next 18 months focusing on adoption, while also hinting at commercial efforts targeting enterprise clients in the near future.
While some may view this strategy as risky, especially if the AI hype wanes, Dottxt is confident in the substance behind the trend. Hugging Face CTO Julien Chaumond, an investor in Dottxt, recently labeled structured generation as “the future of LLMs,” reinforcing the belief that this approach will become increasingly vital.
With the rise of smaller AI models and a growing interest in AI agents, Dottxt’s vision for structured generation could redefine how enterprises harness AI’s potential. “Everyone will be using structured generation in a few years, there is no doubt about that,” Louf predicts. As Dottxt continues to innovate, it might just change the game for developers and AI alike.
I’m a finance writer with three years of experience in investment analysis. At Investorwelcome , I translate complex financial concepts into clear, actionable insights to help investors navigate the market with confidence. Combining my solid academic background with practical industry knowledge, I’m dedicated to providing readers with accurate and timely information. My goal is to empower both new and seasoned investors by simplifying intricate data and offering strategic advice. When I’m not writing, I stay engaged with market trends and investment innovations to ensure my content remains relevant and valuable.