Infrastructure startups in San Francisco Bay Area

Inception

Palo Alto

Inception develops diffusion-based LLMs (dLLMs) that generate answers in parallel, delivering up to 10x faster, more efficient, best-in-class quality. It pioneered diffusion in language, launched the first commercially available dLLMs in 20

Mem0

San Francisco

Mem0 provides a universal, self-improving memory layer for LLM applications, enabling personalised AI experiences while reducing costs and boosting user delight. It markets itself as "The memory layer for Personalized AI."

Browser Use

San Francisco

The company specializes in animating web agents, bringing them to life as living entities on the web.

Runloop

San Francisco

Runloop focuses on infrastructure and orchestration for AI agents, aiming to power the Agentic Web. With a small team of engineers formerly from Google and Stripe, it tackles productionizing AI for software engineering at scale.

Statics

San Francisco

Statics is building Context OS for AI agents to optimize context and improve LLM performance. Backed by Tier 1 VCs, including a16z, it works with a portion of Forbes AI 50 and is an early-stage company led by Johnny Wang.

Feltsense

South San Francisco

Feltsense builds autonomous agents to run entire businesses, aiming to move beyond chat to autonomous management. Backed by $5.1M funding from investors, it operates at the intersection of autonomous systems and applied AI and is scaling in

Mach9

San Francisco

Mach9, founded in 2021 in San Francisco, is a geospatial technology company that automates detection and analysis of urban features for transportation planning, turning geospatial data into actionable insights via machine learning and compu

Maitai

Redwood City

Maitai is an enterprise AI company that manages the LLM stack for enterprises, aiming for the fastest, most reliable inference. It is trusted by YC startups and public companies to manage their models and agents for real-time use.

Morph

San Francisco

MorphLLM develops Fast Apply models for code editing AI, claiming the fastest LLM code-editing inference engine, delivering 10,500 tokens per second per request on Nvidia hardware. Its stack enables high-throughput AI workflows for coding a

Cua

San Francisco

Cua builds infrastructure for general AI agents to safely use computers and apps. It offers an open-source framework and cloud sandboxed container platform, a seed round, 9k+ GitHub stars in four months, and a research-backed blueprint for

Aravolta

San Francisco

Aravolta develops software to monitor, optimize, and control data centers, with a focus on data center infrastructure management (DCIM) software. Its mission is to build the next generation of DCIM solutions.

14.ai

San Francisco

14.ai, founded by Michael Fester and Marie Schneegans, builds customer support solutions for 1000x teams; backed by Y Combinator and SV Angel, with backers including founders from Vercel, Slack, Dropbox, and Algolia, and Fester previously c

Relace

San Francisco

Relace focuses on models and infrastructure for coding agents, defining its core offerings around coding-agent technology.

Relari

San Francisco

Relari is building infrastructure for trustworthy AI agents in the Software 3.0 era, behind Nuvi.dev, and has developed open-source frameworks Agent Contract and continuous-eval to help ensure AI systems behave as intended.

CTGT

San Francisco

CTGT Inc safeguards reputational risk from GenAI for organizations in the B2B software and services infrastructure market. Based in San Francisco as a seven-employee startup founded by Cyril Gorlla and Trevor Tuttle, Gorlla's efficient AI w

Nozomio

San Francisco

Nozomio builds context-augmentation infrastructure for AI agents, indexing codebases, documentation, and external sources to provide agents with expanded knowledge. It claims this boosts performance by 27% and emphasizes fast shipping and r

RunRL

San Francisco

RunRL (X25) is a San Francisco–based reinforcement learning as a service provider delivering on-demand RL capabilities to B2B software and services infrastructure companies, operated by a three-person team led by founders Andrew Gritsevskiy

DeepAware AI

San Francisco

DeepAware AI is a YC-backed startup delivering AI-powered infrastructure management and autonomous robotics for data centers, aiming to improve security, uptime, and energy efficiency by up to 30% for operators and AI infrastructure provide

Pangolin

San Francisco

Pangolin is an open-source proxy and network/ingress orchestration platform that securely exposes applications with identity-aware access control. It has a vibrant community around fosrl/pangolin on GitHub with 15k+ stars, 315k+ deployments

Wafer

San Francisco

Wafer provides a GPU optimization platform for AI model training and inference, plus infrastructure for inference and GPU optimization, focused on optimized AI inference tailored to each stack.

AgentHub

San Francisco

AgentHub is an early seed-stage startup building a simulation and evaluation engine for AI agents, providing the infrastructure layer for agent-driven products. Its team includes the former Apple Foundation Model Evaluation lead and CMU/MIT