Gimlet Labs

Gimlet Labs is an applied AI research and product company that has developed a platform to optimize how AI workloads are run. Founded in 2023 by Zain Asgar, Omid Azizi, Michelle Nguyen, and Natalie Serrino, the San Francisco-based startup spun out of a Stanford research project. The founding team possesses deep experience in AI, distributed systems, and hardware. Zain Asgar, for instance, was previously a GPU architect at NVIDIA and an engineering lead at Google AI.

The company launched in October 2025 with $12 million in seed funding in a round led by Factory, with participation from notable angel investors including the CEO of Intel, the CEO of Figma, and the former CEO of VMware. Upon emerging from stealth, Gimlet Labs was already generating eight-figure revenues from deployments with AI-native and Fortune 500 companies.

Gimlet's core technology decouples agentic AI workloads from the underlying hardware. The platform disaggregates each workload into its component parts and intelligently maps each component to the most suitable hardware, whether it be GPUs, CPUs, or custom accelerators from different vendors like NVIDIA, Intel, and AMD. This process allows developers to deploy their AI applications on the most efficient hardware without rewriting code, which can significantly reduce costs and improve performance. The system features an intelligent workload orchestrator, a hardware-agnostic compiler, and an autonomous kernel generation tool to create optimized kernels for various hardware platforms. Gimlet offers its software stack as both an on-premises solution for data centers and a hosted developer platform, with a serverless inference cloud for AI agents.

Keywords: AI workload optimization, agentic AI, hardware-agnostic compiler, serverless inference, multi-agent systems, GPU efficiency, heterogeneous computing, edge AI, AI infrastructure, model deployment, compute orchestration, kernel generation, Zain Asgar, Omid Azizi, Natalie Serrino, Michelle Nguyen, AI cost reduction, AI model monitoring, datacenter AI, MLOps, AI pipeline