Poolside is an AI-focused private company developing tools for code generation and secure model deployment, and this article provides a weekly recap of its latest product and ecosystem developments. Over the past week, the company advanced both its core Laguna AI model line and its positioning around secure, enterprise-grade AI adoption.
Poolside introduced Laguna XS 2.1, a lightweight upgrade to its earlier Laguna XS.2 model aimed at improving multilingual coding and terminal-style tasks. The new release is designed to be easier to run locally across varied deployment environments, underscoring the company’s emphasis on flexible, on-premise and edge-friendly AI solutions.
The model supports a broad range of frameworks, including vLLM, SGLang, NVIDIA TensorRT-LLM, HF Transformers, and Ollama, which may simplify integration for developers using different infrastructures. Poolside is also offering multiple checkpoint formats such as BF16, FP8, INT4, and NVFP4, giving users options to optimize for performance, cost, and specific hardware configurations.
Alongside Laguna XS 2.1, Poolside released DFlash speculator models that, according to internal tests, can double achieved tokens per second in local inference. This focus on higher throughput and lower latency is likely to appeal to enterprises and developers that prioritize responsiveness, privacy, and reduced dependence on cloud-based inference.
The company has made Laguna XS 2.1 available through multiple channels including Hugging Face, OpenRouter, the Poolside API, Ollama, and pool, its terminal-based coding agent. This multi-platform distribution strategy suggests an effort to broaden adoption and embed Poolside’s technology into existing AI and developer workflows across the ecosystem.
In parallel, Poolside highlighted a session at an Amazon Web Services Summit focused on bringing AI-powered software development into secure environments within an organization’s AWS boundary. Company representative Ivo Pinto is slated to discuss deploying frontier AI models and agents closer to where development occurs, including within more constrained and security-sensitive environments.
The AWS-focused session emphasizes the importance of owning AI models for security, governance, and cost management at scale, aligning Poolside with enterprise and potentially government or regulated-industry customers. By associating its solutions with AWS-centric workflows, the company appears to be deepening ties to mainstream cloud infrastructure where many development teams operate.
Collectively, these developments indicate a week of product enhancement and strategic positioning for Poolside, as it strengthens its Laguna model line, improves local performance, and underscores its fit for secure, enterprise-grade AI deployments.