Announcing Our Investment in dottxt: From Back-and-Forth Prompts to Truly Dependable Computers
At EQT Ventures, we have the privilege of collaborating with visionary entrepreneurs who are shaping the future of AI and its applications. Over the past few years, we’ve closely followed the rapid evolution of Large Language Models (LLMs) and their transformative impact across industries. Yet, one pressing challenge has consistently emerged: integrating LLMs seamlessly into existing digital ecosystems.
LLMs, while powerful, are inherently probabilistic and often produce unpredictable outputs. This lack of structure hinders their integration into applications where precision and reliability are paramount. Imagine if we could guide LLMs to generate outputs in specific, structured formats — turning them from impressive conversationalists into dependable components of our software systems.
Meet dottxt: an AI developer tool that makes LLMs speak the language of every application, unlocking the key to building AI apps at scale.
dottxt has developed an innovative methodology to shape LLM outputs into widely used data formats, before any text is generated. All without compromising on speed. This preemptive structuring is a game-changer, elevating LLMs from simple chat functions to dependable computers that can be seamlessly integrated into existing systems.
We’ve been excited by:
1. The Team: From our first interactions, we’ve been impressed by Rémi, Dan, and Brandon’s brilliant minds and ambition. Their unique backgrounds in statistical modeling and compiler technology, combined with extensive experience at AI leaders like Hugging Face and Normal Computing, gives them a unique position to tackle this challenge. Over the past four years, they’ve honed their collaboration, bringing together a wealth of knowledge and a shared vision.
2. The Timing: While they’re coming out of stealth today, their open-source library, Outlines, has already been downloaded over 3 million times and is used by industry leaders like OpenAI and Cohere. The feedback from our AI community has been unprecedented and unanimous: “If they solve this, everyone will want to use it.”
3. The Unique Positioning of the Product: dottxt takes a radically different approach by enabling LLMs to produce structured, reliable outputs that can be directly integrated into applications. This elevates LLMs from probabilistic models to tools that can be programmed with precision, unlocking their full potential. By implementing guardrails based on unique criteria, dottxt effectively stops LLMs from making errors before they even make them.
It’s always exciting to find a team that has built the right product at the right time. dottxt is providing the technological capabilities that developers and enterprises are eagerly seeking.
We believe that by making AI programmable and enabling LLMs to produce structured, reliable outputs,dottxt is spearheading a new paradigm in the AI industry.
Today, we’re proud to be leading their $11.9 million funding round, bringing on board an experienced group of founders and operators like Erik Bernhardsson (CEO of Modal Labs), Julien Chaumond (CTO of Hugging Face), Bob van Luijt (CEO of Weaviate).
A warm welcome to Rémi, Dan, Brandon, and the entire dottxt team to EQT Ventures!
dottxt is currently hiring. https://jobs.ashbyhq.com/dottxt if you’re curious about joining as an early team member of one of the most exciting AI companies in the world.
At EQT Ventures, we are actively investing in AI, LLMs, and the technologies that unlock their full potential. If you are building in this space, please reach out to julien.hobeika@eqtventures.com
— -
Imagine waking up one morning as an entrepreneur in the AI space to find OpenAI and its CEO Sam Altman crediting your code as a source of inspiration. For Rémi, Dan, and Brandon, the creators behind dottxt, this isn’t a fantasy — it’s their reality.
