Yavqo Labs was created around a simple idea: some of the most important work in artificial intelligence requires a dedicated place to think beyond the next product cycle.
As AI became more central to Yavqo, we saw the need for a team focused not only on applying existing technology, but on developing new models, new systems, and new ways for people to interact with software.
That became Yavqo Labs.
Building for the long term
Product teams are often optimized around clear roadmaps, customer needs, and release timelines. Research works differently.
Progress can come from testing ideas that may not immediately become features, improving core capabilities that take years to mature, or exploring technical directions whose value only becomes clear over time.
Yavqo Labs gives those efforts a dedicated home.
The organization brings together researchers, engineers, and product teams working across areas including:
- Foundation models
- Multimodal AI
- Real-time voice and conversational systems
- AI agents
- Model efficiency and inference
- Safety and reliability
- Developer infrastructure
- New human-computer interfaces
The goal is not research for its own sake. It is to create technology that can eventually make Yavqo products more capable, useful, and accessible.
From research to products
One of the principles behind Yavqo Labs is maintaining a close connection between research and the products people actually use.
That means new capabilities can move from experimentation into real systems more quickly, while product teams can bring practical challenges back into the research process.
Sunlight is one example of that approach.
Work across language models, conversational AI, agents, and real-time systems has helped shape the Sunlight model family and the products built around it.
The same approach extends beyond AI assistants. Research from Yavqo Labs can contribute to browsers, mapping, infrastructure, devices, developer tools, and other parts of the Yavqo ecosystem.
A dedicated home for AI research
Creating Yavqo Labs also gives us the ability to invest in problems that are broader than any single product.
Some of those problems are technical, such as reducing latency, improving reasoning, making models more efficient, or building systems that can operate reliably at scale.
Others involve understanding how increasingly capable AI should interact with people, developers, and the systems around it.
These are not problems that can be solved by one team or one release.
They require sustained work.
Working across Yavqo
Yavqo Labs is designed to work closely with teams throughout the company.
Researchers collaborate with infrastructure engineers on training and inference systems. Product teams help identify where new capabilities can create meaningful improvements. Safety and security teams work alongside model developers as systems become more capable.
This structure allows research to remain ambitious without becoming disconnected from the realities of building products used at scale.
What comes next
Yavqo Labs is still evolving.
As our models become more capable, we expect the organization to expand into new areas of AI research and to work more deeply across Yavqo's products and platforms.
Some projects will become products quickly. Others may take years. And some may simply help us understand what should be built next.
That long-term perspective is exactly why Yavqo Labs exists.
We built it to create a place where ambitious AI research can become useful technology, and where the ideas shaping the future of Yavqo can begin.
