Poolside is a foundation model company committed to driving human-level artificial intelligence by focusing on software engineering as a strategic beachhead. Founded in 2023, the organization believes that software development, which requires complex reasoning, planning, and understanding across long-horizon objectives, serves as the most effective route to achieving artificial general intelligence. By creating robust foundation models and agentic systems, Poolside aims to deliver intelligence into the environments where actual work occurs, ranging from local workstations to high-consequence enterprise and government infrastructure.

The core of the Poolside platform is a comprehensive stack consisting of foundation models, developer tools, and agent orchestration. These components are designed to function within a client's specific security boundaries, ensuring that sensitive data, proprietary code, and operational workflows remain private and under user control. The platform distinguishes itself by offering on-premises, VPC, and air-gapped deployment options, avoiding reliance on external cloud-only APIs. Furthermore, Poolside employs a team of Forward Deployed Research Engineers who work directly alongside client teams to ensure the integration of AI capabilities translates into measurable business outcomes, such as improved system reliability and reduced delivery timelines.

Some of the key features are:

• Laguna Models: Specialized foundation models trained from scratch for agentic coding, including the lightweight Laguna XS.2 and the highly capable Laguna M.1.
• Model Factory: An internal end-to-end orchestration platform that automates evaluation, synthetic data generation, and reinforcement learning during the training process.
• Agentic Orchestration: Systems that enable agents to plan, use tools, and execute code in sandboxed environments while adhering to strict policy controls and auditability requirements.
• Native Developer Integrations: Support for IDEs like VS Code, Visual Studio, Zed, and JetBrains, allowing for a seamless experience without forcing context-switching.
• Enterprise Governance: Centralized management through the Poolside Console, providing visibility into agent sessions, version control for AI-assisted work, and fine-grained role-based access control.
• Secure Deployment: Options for air-gapped, STIG-hardened, and disconnected environments, specifically optimized for sensitive workloads in the public sector and government missions.

The system operates by embedding AI-native agents into the existing software development lifecycle. These agents function as assistants that can interact with file systems, run commands, and access APIs within predefined permission scopes. Because the platform follows an open, modular design—utilizing standard protocols like the Agent Client Protocol—it can be integrated into diverse environments, from large-scale enterprise clouds to compact local towers, all while maintaining full end-to-end traceability and audit logs for compliance.

Some common use cases include:

• Secure Software Development: Empowering developers to use AI-assisted coding and agentic workflows within strictly air-gapped or classified networks.
• Automated CI/CD Workflows: Integrating autonomous agents into pipelines to automate code reviews, testing, and generation on every push.
• Enterprise Agentic Transformation: Mobilizing multi-agent systems to manage complex, multi-step engineering tasks while maintaining corporate-grade security and secret management.
• Government Mission Support: Providing high-consequence intelligence for national security tasks where data sovereignty and infrastructure independence are non-negotiable.