Poolside Highlights Laguna Open-Weight AI Models in Enterprise-Focused Weekly Developments

Poolside advanced its strategy in enterprise-focused AI this week, spotlighting its new Laguna family of agentic coding models and their potential role in real-world software development workflows. The company framed these releases as part of a broader industry shift from AI experimentation to large-scale deployment in mission-critical environments.

Poolside formally introduced Laguna M.1 and Laguna XS.2, its first public in-house models designed for long-horizon coding tasks such as writing code, running tests, and debugging failures. Laguna M.1 is a 225B-parameter Mixture-of-Experts model with 23B activated parameters trained on 30T tokens, while Laguna XS.2 is a 33B-parameter MoE model with 3B activated parameters released under the Apache 2.0 license.

The company underscored Laguna XS.2 as its first open-weight model, positioning open weights as central to its enterprise strategy rather than a side initiative. Poolside highlighted benefits such as greater control, cost visibility, and deployment flexibility for enterprises, aiming to serve compliance-sensitive customers that need customizable, internally deployable AI systems.

Poolside said the Laguna models emerge from an internal “model factory” developed over several years, including proprietary data pipelines, a dedicated training stack, and agent infrastructure. The firm also cited deployment experience in demanding enterprise and government settings, using this track record to signal production maturity and reliability to potential buyers.

To expand distribution, Poolside is making Laguna accessible via its own API, an agent harness called pool, and a browser-based environment named Shimmer. Partnerships with Hugging Face, OpenRouter, Baseten, and Ollama are intended to lower friction for developers and accelerate ecosystem adoption, creating more avenues for usage and feedback.

The company emphasized a vision where successful models form a “substrate” for others to build upon, aiming to foster an ecosystem around Laguna that could strengthen network effects and increase switching costs. With a roughly 60-person team spanning applied research, infrastructure, and product, Poolside appears focused on a vertically integrated approach that ties model development closely to real-world deployment.

From a financial perspective, the combination of open-weight releases, enterprise-oriented capabilities, and broader distribution could enhance Poolside’s market reach and brand visibility in AI infrastructure and developer tools. If enterprises validate the Laguna models for long-horizon coding and mission-critical use cases, these steps may support a more durable competitive position and pave the way for scaling recurring revenue over time.

Overall, it was a pivotal week for Poolside as it transitioned its core AI models into the public domain, sharpened its enterprise narrative around open weights and governance, and laid groundwork for a broader developer ecosystem around agentic coding.