Infrastructure Startups funded by Y Combinator (YC) 2026

November 2026

Browse 300 of the top Infrastructure startups funded by Y Combinator.

We also have a Startup Directory where you can search through over 5,000 companies.

  • Fivetran
    Fivetran
    Y Combinator LogoW2013
    Active • 1,200 employees • San Francisco
    Fivetran automates data movement out of, into and across cloud data platforms. We automate the most time-consuming parts of the ELT process from extracts to schema drift handling to transformations, so data engineers can focus on higher-impact projects with total pipeline peace of mind. With 99.9% uptime and self-healing pipelines, Fivetran enables hundreds of leading brands across the globe, including Autodesk, Conagra Brands, JetBlue, Lionsgate, Morgan Stanley, and Ziff Davis, to accelerate data-driven decisions and drive business growth. Fivetran is headquartered in Oakland, California, with offices around the world. 
    data-engineering
    saas
    analytics
    b2b
  • Orikata Bio PBC
    Orikata Bio PBC
    Y Combinator LogoF2026
    Active • 1 employees • San Francisco
    A new agent, tool, or generative model alone doesn’t solve the whole problem for most research questions. Someone still has to assemble the data, work out which methods to use and get the tools and datasets working together. We’re building Noto, a multiplayer discovery workbench for human and agentic scientists, to bring everyone's favorite frontier models, agents, tools, and scientific models into one shared workspace, so that you can focus on making your next discovery, rather than wrestling with your Python environment or finding GPUs. Plan an experiment, work with its generated data, run an analysis or model, maybe even train a new model, and keep the results in the notebook. We designed Noto so that the inputs, code, methods and results stay connected across local and hosted compute. Both people and agents need to see how a result was produced and build on it. Biology is where we’re starting, with the workbench being built for more general research across many scientific disciplines. Available now: our first biological analysis tools, hosted model inference, and an initial agent interface. Free usage is available within limits, alongside paid model runs and individual and team plans. Researchers can run sequence and structure analyses and hosted scientific models without provisioning GPUs or maintaining each model’s dependencies. In development: the full multiplayer notebook and wider agent workflows and integrations. Orikata Bio Public Benefit Corporation, or Orikata Bio PBC, is the company. We work under two brands: Orikata Bio for biological discovery and Megure Labs for scientific software. Biological discovery with Orikata Bio We’re developing Ibuki, a world-model family for biology, to learn how evolution, sequence, structure and function relate and use those relationships to understand and engineer living systems. We’re pretraining our own protein language models, and developing coordinate and multimodal models for prediction, inverse folding, alignment and generative design. Proteins are our starting point; DNA, RNA, small molecules, and other biological modalities are part of the longer-term direction. Along the way, we assembled Umi, with over 100 billion unique protein sequences, and Sora, with nearly two billion protein structure records. Having that much data only helps if we can organize it, compare it and find the biology relevant to a particular question. We’re building the full-corpus embedding, indexing and customer access that will bring more of those collections into research workflows. Our key insight was that the fixed substitution tables used to compare biological sequences are the simplest example of a neural network. Replacing them with contextual substitution operators lets us learn how positions correspond, from reference alignments and directly from molecular geometry. Hikoboshi (open source) turns those learned relationships into tools for sequence and structure alignment, search, clustering and many other evolutionary analysis pipelines. GitHub: https://github.com/orikata-bio/hikoboshi Scientific software with Megure Labs Learning those relationships meant making the alignment calculation itself trainable. Orihime (open source), our differentiable-alignment and structured-attention library, lets neural networks learn through scientific algorithms on CPUs and GPUs. It supplies the training machinery behind our learned alignment methods. GitHub: https://github.com/megure-labs/orihime Running the models efficiently is its own problem. Hataya (private evaluation), our inference engine and compiler for scientific models, turns trained models and their dependencies into standalone CPU and GPU programs, carrying the exact weights, identity and verification evidence with them. We’re building it to make both our models and public scientific models cheaper to run, including through Noto’s hosted service. Kaname coordinates long-running agent work across every model provider: tasks, compute, permissions, retries and the record of what happened. We’re bringing that coordination and provenance into Noto’s research workflows so work can continue across people, agents and machines. Researchers can start with a single analysis or hosted model run today. Noto is where we’re bringing them together so an analysis or model run can become part of a continuing research project, with the data, methods and results kept alongside it. Try Noto’s live tools: https://megure.ai/noto/workspace/ Biological tools: https://orikata.ai/tools/ Scientific software: https://megure.ai/ Biological discovery: https://orikata.ai/ Open source: https://github.com/megure-labs and https://github.com/orikata-bio For research use, questions about the data or private evaluations: casey@orikata.ai
    saas
    enterprise-software
    machine-learning
    ai
  • Ample Computer
    Ample Computer
    Y Combinator LogoF2026
    Active • 2 employees • San Francisco
    Vibecoding has made creating software easy, but launching a real app still means wiring up 5-10 separate cloud services, which stops most new builders. Ample is the one-stop cloud for agents to ship complex web apps. Add Ample to Claude or ChatGPT, tell your agent to "deploy on Ample", and get a live URL in seconds. Ample reads your codebase, works out what infrastructure it needs, and sets up compute, database, storage, CDN, domain and email, with payments, text and analytics next. We then monitor, patch and scale your app so you never have to migrate it. Ravi and Ishan led product and infrastructure engineering for the Shopify App Store, a platform of 25,000 apps used by 5 million merchants and processing $1B in GMV a year. We're doing for software what Shopify did for e-commerce, so that anyone with a computer or even a phone can deploy, share and sell ambitious software from any AI platform.
    b2b
    cloud-computing
    infrastructure
    ai
  • Reific
    Reific
    Y Combinator LogoF2026
    Active • 1 employees • San Francisco
    Reific builds RL environments for superhuman reasoning over time, for the frontier labs that train agents. AI has learned to reason over text and code. But most of what matters in the real world shows up as numbers that change every second: a heartbeat, a gearbox, a GPU cluster. No human can track thousands of signals at once, but AI should be able to. In our environments, data streams in on a clock while the agent works, and every action is graded by what actually happened. Most AI training today teaches models to match human experts. We treat human performance as the floor. Each environment comes from real, de-identified records, in three task families: personal agents, industry, and science & compute. Today, even the best frontier model averages under 50% on each family. Training an open model on our environments raised its full-solve rate on unseen tasks by 54% (from 25.6% to 39.5%), with no loss on general benchmarks. Our mission is to build the infrastructure that keeps AI climbing the frontier, long past human level in the real world.
    reinforcement-learning
    machine-learning
    generative-ai
    infrastructure
    ai
  • Limnion
    Limnion
    Y Combinator LogoF2026
    Active • 3 employees • Seattle
    Limnion is building an open-standards telemetry storage platform for observability at scale, including agentic workflows. Siglake is the open-source project that sits at the heart of it, designed around OTel, Parquet, and Iceberg. No proprietary storage, no vendor lock-in.
    databases
    open-source
    infrastructure
  • Routant
    Routant
    Y Combinator LogoF2026
    Active0 • San Francisco
    AI agents won’t stay inside the companies that build them. The most valuable agents will work across company boundaries, inside customers’ private systems. In those environments, intelligence is no longer the barrier to deployment. Trust and access are. Routant is the zero-trust access layer for this new agentic world. Vendors integrate once, while customers retain governance over every deployment, deciding what agents can access and which actions need human review, with a complete audit trail of every action.
    cybersecurity
    aiops
    ai
  • Superserve
    Superserve
    Y Combinator LogoF2026
    Active • 3 employees • San Francisco
    Superserve gives every AI agent its own computer in the cloud. Each one is an isolated VM that starts in under 50ms, sleeps between turns, and picks up exactly where it left off, for hours, weeks or indefinitely. Traditional cloud infrastructure was built to run one app for many users. Agents flip that. Each agent needs its own isolated machine where it can write and run code, use tools and work on its own for long stretches. Everything it produces has to persist the whole time, the way your laptop keeps your work. That means agent computers have to be stateful, secure and cheap to run. For agents that talk to customers, they also have to start, sleep and wake almost instantly. Fast - Machines start in under 50ms, either cold or from a template with your repo and dependencies already loaded. You don't need to keep a pool of servers running and waiting. Persistent - Paused machines keep their full memory and disk state at zero compute cost. They can resume or fork from that exact state. Pausing and resuming is fast enough that a machine can sleep between inference calls and wake on the next turn. Agents can run Docker, databases, browsers or any other service inside. Secure - Every machine is a microVM with its own kernel. You can program the rules for what traffic can come in and go out, and they're enforced at the network boundary where the agent can't reach them. API secrets are handled at the network layer too, so they never enter the machine. Every action is logged for audits. Teams building coding agents, autonomous workers, and eval or RL training environments use Superserve to give every agent session or rollout its own machine, with no infrastructure to manage. We run 100K+ machines every week for teams including Lindy and Y Combinator.
    infrastructure
    cloud-computing
    ai
  • Tesser
    Tesser
    Y Combinator LogoW2027
    Active0
    Tesser is a cloud development runtime that lets developers run multiple dev environments in parallel without overloading their laptops. Our CLI moves workloads like servers, builds, tests, and typechecking to isolated cloud machines while developers continue using their existing editors and coding agents. Tesser handles provisioning, networking, and port forwarding. Teams can access each other's environments through our gateway, share live previews, and run Tesser on our infrastructure or their own cloud (BYOC).
    developer-tools
  • FlashDrive
    FlashDrive
    Y Combinator LogoF2026
    Active • 2 employees • San Francisco
    Our mission is to rebuild the file system, the oldest idea in computing, for its newest users: AI agents. FlashDrive is a shared drive for teams and their AI agents. You can use it via MCP or mount it like a normal folder on macOS and Linux, so your laptop, your teammates' and every sandbox work in the same live files at once. Agents hand off work through files: one writes, the next one reads. That makes AI multiplayer across a whole company. Everyone keeps the agent they already like, whether that's Claude Code, Codex, Cursor or something custom, and there's no new app to learn. Your agents and your teammates' agents work in the same files and pick up where each other left off. Shared drives were built for people. Agents work differently: many of them edit the same files at once. Unlike Dropbox or Google Drive, FlashDrive merges edits made at the same time, or keeps both, instead of creating a conflicted copy. Cloud sandboxes connect with no setup, huge files stream instead of filling the disk, and every file can be searched with keywords and grep. (FlashDrive is by Evokoa, Inc.)
    developer-tools
    infrastructure
    databases
    cloud-computing
    ai
  • Orgo
    Orgo
    Y Combinator LogoF2026
    Active0 • San Francisco
    Orgo is the simplest way to give your AI employees their own secure, persistent cloud computers. Over 1000 teams use Orgo to run AI workers.
    cloud-computing
    developer-tools
    ai
  • ObsessionDB
    ObsessionDB
    Y Combinator LogoF2026
    Active • 5 employees • San Francisco
    ObsessionDB is the fastest and most scalable data warehouse for AI. You can process trillions of rows in milliseconds. Companies use it for agent analytics, agentic search, real-time APIs, and more. We offer 10-100X better performance at a much better price.
    big-data
    analytics
    databases
  • Tilo
    Tilo
    Y Combinator LogoF2026
    Active • 2 employees • San Francisco
    Tilo gives AI agents one shared computer. Today every agent gets its own virtual machine and works by screenshots and clicks, which is slow, expensive and brittle. Agents that work together have to describe what they did to each other in messages. Tilo is a Linux-based operating system built for agents as users. Every app exports typed actions that agents call over MCP while people use the same actions as buttons, on the same live state. Many agents work in one workspace at once and read each other's work instead of messaging about it. Every change is attributed and can be undone, and irreversible actions need permission. Because state and interface are separate, agents can generate an app's entire UI on demand. On the same multi-app task with the same model, agents on Tilo were about 29× cheaper and 90× faster than computer use, which failed. A small model on Tilo did it for about 1/10,000 of the frontier model's cost. In our scale test, 50 agents worked in one shared workspace on a single laptop, at about 60 MB of memory per agent. Tilo runs headless, in our cloud, or as a desktop, and you can open any workspace from the browser.
    developer-tools
    infrastructure
    open-source
    ai
  • Pomerado
    Pomerado
    Y Combinator LogoF2026
    Active • 2 employees • San Francisco
    Pomerado turns any website into an API / MCP in minutes. Connect your agents to legacy portals, consumer websites, and internal tools. Why Pomerado: • 7x faster than browser agents on customer workflows • 100x cheaper than browser agents • Self healing, deterministic APIs • Universal website coverage • Hosting, authentication, and permissions handled Customers can use us to: • automate back office workflows on legacy systems (ERP, EHR, CRM) • speed up customer integrations for forward deployments • connect enterprise assistants to internal tools • supercharge personal assistants with fast and reliable browser use We’re ex-McKinsey and Scale AI. We drove enterprise agent deployments at F500s and orgs like the Mayo Clinic. We saw how legacy systems block agent development and AI transformations; Pomerado turns those systems into AI-native components automatically.
    ai
    developer-tools
    api
    b2b
    workflow-automation
  • Acyclic Labs
    Acyclic Labs
    Y Combinator LogoF2026
    Active • 2 employees • London
    We’re building the foundations of agent swarms that can materially accelerate all work: benchmarks to measure model capabilities, models trained to coordinate thousands of agents, RL environments to improve them, and infrastructure that can put a datacenter’s worth of compute behind a single conversation within seconds. Both founders graduated top 10% from Cambridge. Ram is a published ML researcher and Varun worked at the frontier of parallel programming languages and shipped enterprise multi-agent systems at Palantir and Tracelight. We realised that agents were good enough to solve most use cases. So why couldn’t we run thousands of them at once and finish hours of work in minutes?
    ai
    infrastructure
    data-engineering
    swarm-ai
    open-source
  • Stacktrace
    Stacktrace
    Y Combinator LogoF2026
    Active • 3 employees • San Francisco
    Agents will do most of the work in the future, but only if teams can trust them unsupervised. Stacktrace is the control layer for every agent: it sees every action, detects risky behavior, enforces the bounds your team sets, and steps in when an agent drifts from intent.
    infrastructure
    devsecops
    cybersecurity
    ai
  • Terse
    Terse
    Y Combinator LogoF2026
    Active • 2 employees • New York City
    We build infrastructure to power agentic systems such as software factories, multiplayer agent frameworks and agent coordination.
    infrastructure
    ai
    api
  • GrayPass
    GrayPass
    Y Combinator LogoF2026
    Active • 2 employees • Toronto, ON, Canada
    Login tells you who signed in. GrayPass helps apps check who’s behind sensitive actions and whether they’re authorized before those actions go through. It combines identity, device and session context with behavioral signals like typing rhythm, cursor movement and scrolling. Before a sensitive action goes through, GrayPass evaluates that evidence against your team’s policy and returns a decision: allow, challenge, hold or deny. Your app enforces the decision. This applies to actions like exporting sensitive data, changing admin permissions or allowing an AI agent to act on someone’s behalf. Allowed actions receive short-lived, signed proof tied to the exact action and your app. Your existing login and authentication stack stays in place.
    identity
    biometrics
    infrastructure
  • Agent 37
    Agent 37
    Y Combinator LogoF2026
    Active0 • San Francisco
    Agent 37 builds the cheapest sandboxes for running AI agents. A sandbox is the cloud computer an agent runs in. Every agent needs its own computer to keep its files and run code. Agents are going to vastly outnumber humans, and sandboxes today are too expensive to run at that scale. Ours cost 94% less than most sandbox providers, and you pay by the minute, for the time your agents run.
    developer-tools
    saas
    b2b
    ai
  • Tilion
    Tilion
    Y Combinator LogoF2026
    Active • 4 employees • San Francisco
    we build self-healing browser infrastructure that runs millions of agent operations a day without supervision. teams use it for large-scale data collection, agentic workflows, continuous monitoring, and anything else that requires acting on the web at volume.
    b2b
    infrastructure
  • Hopper
    Hopper
    Y Combinator LogoF2026
    Active • 2 employees • San Francisco
    We are building the inference layer for voice. Voice agents are latency-sensitive and differentiated enough to warrant a focused inference player, similar to Fal (inference for video) or Sail Research (inference for long horizon agents). Before this our team wrote megakernels at Meta Reality Labs and ran inference at Salient (a16z-backed voice startup)
    infrastructure
    ai
  • Maritime
    Maritime
    Y Combinator LogoF2026
    Active • 3 employees • San Francisco
    Maritime is a fully managed cloud platform for startups selling AI agents. We help them run thousands of agents, each in its own isolated, persistent computer, starting at $1 per agent per month, with all underlying infrastructure, scaling, sleep/wake, and monitoring handled automatically.
    developer-tools
    infrastructure
    cloud-computing
  • Boat
    Boat
    Y Combinator LogoF2026
    Active • 3 employees • Paris, France
    **Problem** Humans used to need laptops to do their work. Now they prompt agents instead, and those agents need computers too: always-on, in large quantities, and with unlimited capabilities. Building and maintaining that yourself is far from trivial. **Solution** Boat provides powerful, simple, permissive Ubuntu VMs in the cloud. In a sandbox, an agent can work for long periods on code or on knowledge work like legal, research, reporting, and auditing. **Why Boat and not any other cloud VM?** 1. It is ready out of the box: every sandbox ships with a desktop, chrome, languages, docker, snapshots, and templates, plus a CLI, SDK, API, and control pane for your fleet. 2. It is permissive and paradigm agnostic: any number of agents can use any number of sandboxes. Agents can run inside or use sandboxes as an external tool. 3. It is expertly crafted: we manage everything from the individual CPU to the orchestration software, which makes sandboxes stable, persistent, scalable, and secure. 4. It is the best performance and price in town: our CPUs rank first on HPC benchmarks, and sandboxes are typically 9x to 18x cheaper than other leading sandboxes. **Further notes** Sandboxes are hosted in Europe, which keeps prices low, and more regions are coming. Unless compliance requires it, agents don't notice the region they run in.
    developer-tools
    cloud-computing
  • Tokenless
    Tokenless
    Y Combinator LogoS2026
    Active • 3 employees • San Francisco
    Tokenless is an applied AI research lab. Our goal is to end reward hacking. We do research and build tools to detect issues with environments and evals, understand model behavior in those environments, and prevent those issues from happening. Correcting them makes training more effective, makes model performance easier to interpret, and is crucial to solving the AI alignment problem. If you’re training models with RL, building environments, or running agent evals, and want more confidence in your environments, we're here to help.
    ai
    b2b
    machine-learning
    reinforcement-learning
    developer-tools
  • Magma
    Magma
    Y Combinator LogoS2026
    Active • 1 employees • San Francisco
    We enable companies monetize their agents' traces.
    ai
    reinforcement-learning
    data-engineering
  • Riften
    Riften
    Y Combinator LogoS2026
    Active • 5 employees • San Francisco
    Riften is an independent AI research lab exploring intelligence in software and the physical world. We study how models learn and represent knowledge, and build systems that put that research to work. Our work spans model routing, evaluations, post-training, and physical AI. Earned intelligence.
    developer-tools
    reinforcement-learning
    infrastructure
    ai
  • Caution
    Caution
    Y Combinator LogoS2026
    Active • 5 employees • George Town, Cayman Islands
    Caution is a hosting platform for software you can't afford to get hacked. It simplifies and extends confidential compute, and lets developers deploy workloads to secure enclaves in minutes and give customers, auditors, and counterparties cryptographic proof that production is running the exact reviewed source code, build, configuration, and runtime - not an opaque binary controlled by a cloud operator or internal admin. Confidential computing hardware exists, but the tooling is too fragmented and incomplete for most teams to use well. Today, attestation often proves only that a binary has a certain hash; it does not prove what source code produced that binary. Building the missing deployment, reproducibility, and verification layer requires specialized security engineering and often locks teams into one cloud or hardware stack. Caution turns that into a simple git-based workflow that builds reproducible enclave images, provisions infrastructure, and lets anyone independently verify what is running. We built Caution after years securing high-risk systems through Distrust, where we worked with hedge funds, custodians, blockchain teams, cloud platforms, and critical grid operators on systems responsible for more than $600B in assets. Again and again, the same problem showed up: teams running mission-critical infrastructure could not reliably prove what was executing in production. As AI, fintech, healthcare, and regulated data collaboration move into the cloud, "trust us" is no longer enough. Caution is a general-purpose platform: if it runs on a server, it can run verifiably on Caution. We're starting where trust failures cost the most, but we believe verifiable compute will become the default way software runs, the same way HTTPS became the default way it's served. Infrastructure you can verify, not trust.
    infrastructure
    cloud-computing
    cybersecurity
  • OpenVector
    OpenVector
    Y Combinator LogoS2026
    Active • 4 employees • San Francisco
    OpenVector transforms any existing camera into an automated 24/7 workforce without requiring new hardware. We've built the world's fastest vision-language model inference engine that allows users to simply describe a task in plain English and automatically build a custom AI vision model to monitor their feed. When it detects an important physical event—like a stalled production cycle, a safety violation, or a waiting customer—it operates a computer just like a human to take immediate action. It integrates with legacy systems, makes calls, sends emails, updates a POS/CRM, or triggers webhooks in real time. OpenVector enables any camera to do real work.
    computer-vision
    robotics
    infrastructure
    industrial
  • Autoresearch, Inc.
    Autoresearch, Inc.
    Y Combinator LogoS2026
    Active • 3 employees • San Francisco
    Autoresearch helps teams train agents for tasks where general models are not good enough. Give us a task, your current agent or model, and an evaluation. Autoresearch runs an autoresearch campaign across data, SFT and other post-training methods, continual learning, prompts, tools, harnesses, and systems, then returns the best verified improvement. It runs hosted or in your cloud. We provide the research system around models: persistent memory, full experiment lineage, compute orchestration, and statistical verification. We’re starting with agent and model training, continual learning, and optimization. Reach out to us for early access at founders@autoresearch.inc.
    developer-tools
    infrastructure
    machine-learning
    reinforcement-learning
    ai
  • Executor
    Executor
    Y Combinator LogoS2026
    Active • 1 employees • San Francisco
    Executor.sh is an MCP gateway that allows you to sign in once to all of your integrations and use them with any client. We enable you to connect your agents to anything (MCP, GraphQL, OpenAPI) so regardless of what you're trying to connect agents to you're able to. If you're trying to scale agents to access all of your production services, data, etc, you need Executor as a unified management layer.
    ai
    aiops
    saas
  • GitCafe
    GitCafe
    Y Combinator LogoS2026
    Active • 2 employees • San Francisco
    We build GitHub for agents. We're rebuilding the infrastructure for the next hundred billion commits and CI jobs with hundreds of engineers already making the move.
    devops
    developer-tools
    ai
    b2b
  • Decawork
    Decawork
    Y Combinator LogoS2026
    Active • 2 employees • San Francisco
    An agent’s access should depend on the task it is doing. The same action can be appropriate for one task and wrong for another. Access decisions need to adapt in real time as the agent works, but companies can’t turn every business judgment into a fixed rules or have people manually approve every step. We train small, company-specific models that approve AI actions in real time. They consider the company’s policies and interests, the agent’s current task, and its past actions to decide whether each proposed action should be allowed. Access is specific to the task at hand and reassessed as the agent works. Companies will rent intelligence from AI labs or run open-source models to power their agents. They’ll own the models that keep their entire AI workforce acting in their interests. Decawork is building those models.
    b2b
    infrastructure
    security
    ai
  • Orchestra
    Orchestra
    Y Combinator LogoS2026
    Active • 2 employees • San Francisco
    Orchestra is an inference cloud. We capture your AI work traces, train smaller models, and automatically deploy them (only when they outperform the expensive model you are currently paying for). Instead of renting the same frontier model forever, your AI infrastructure gets faster, smarter, and cheaper every time it runs. Suppose your product uses Claude for a repetitive legal workflow. Orchestra observes successful production runs, learns the specific tools and judgment required for that workflow, and trains a specialized open-weight model. Once that model matches or beats Claude on a held-out evaluation, Orchestra starts routing production traffic to it. As more work is completed, the model keeps improving. We started Orchestra because we believe every company using AI should be accumulating intelligence, not accumulating API bills.
    ai
    ml
    open-source
  • Speko
    Speko
    Y Combinator LogoS2026
    Active • 4 employees • San Francisco
    Speko is a router for voice models: one API key, and your voice agent runs on the speech-to-text, LLM, and text-to-speech models we measured as best for your language and use case. The measurements are public at benchmarks.speko.ai: we test new models the day they ship and re-route when the numbers change, so you rewrite nothing. We ran this bake-off by hand for four years at previous company, shipping production voice for Fortune 500s.
    developer-tools
    conversational-ai
    ai
  • OneTriangle
    OneTriangle
    Y Combinator LogoS2026
    Active • 4 employees • San Francisco
    OneTriangle optimizes prefill, speeding up TTFT 10x utilizing KV cache transfer tech We integrate with your dedicated endpoints, or GPUs to speed up inference.
    open-source
    infrastructure
    ai
  • Inkbox
    Inkbox
    Y Combinator LogoS2026
    Active • 3 employees • San Francisco
    Give your agents a computer to run on, with their own email, phone, iMessage, and WhatsApp. Each agent runs next to its communication channels, so it never misses a reply, and sleeps when idle, so customers don’t pay for downtime.
    developer-tools
    infrastructure
    ai
  • Edviro
    Edviro
    Y Combinator LogoS2026
    Active • 2 employees • San Francisco
    Edviro builds AI that operates energy infrastructure. We connect to meters, HVAC systems, batteries, solar, and other energy equipment and continuously handle the work required to keep them efficient: detecting and diagnosing problems to forecasting demand, simulating fixes, creating work orders, coordinating maintenance, and controlling equipment where authorized. Our energy world models learn how each site behaves, choose the highest-value actions, and verify the results against real operational, meter, and billing data.
    deep-learning
    b2b
    energy
  • Experiential Labs
    Experiential Labs
    Y Combinator LogoS2026
    Active • 2 employees • San Francisco
    We are building an open source gateway for AI budget holders to control model access and turn their traffic into better models that they own. We're a team that has earned an ML PhD at Georgia Tech, scaled AutoGPT to 160k GitHub stars, and led a self-driving simulation team at Waabi.
    reinforcement-learning
    infrastructure
    ai
  • Duet Machines
    Duet Machines
    Y Combinator LogoS2026
    Active • 3 employees • San Francisco
    We post-train small open models on your AI agent's production traffic, prove them against your experts' graders, and roll them out gate by gate to replace frontier APIs at a fraction of the cost.
    machine-learning
    reinforcement-learning
    infrastructure
    ai
    b2b
  • Parasma
    Parasma
    Y Combinator LogoS2026
    Active • 2 employees • San Francisco
    We train human brain cells for compute
    machine-learning
    neurotechnology
    ai
  • Markov
    Markov
    Y Combinator LogoS2026
    Active • 2 employees • San Francisco
    Markov is the data lab for engineering and computer-use AI. We capture real work from domain experts and craft high-fidelity datasets and RL environments. We built the most complex tasks and RL environments sourced through manufacturing experts proficient in AutoCAD, SOLIDWORKS, CATIA, Revit, Siemens NX and more. Our computer-use datasets are action-labelled, narrated and sourced from domain experts in design and enterprise software. We work with frontier AI labs and we've built over 10,000 CAD environments and collected over 500,000 hours of computer-use data. Our open-source datasets have 800,000+ downloads on Hugging Face.
    reinforcement-learning
    data-labeling
    data-engineering
  • Mireye
    Mireye
    Y Combinator LogoS2026
    Active • 2 employees • San Francisco
    AI agents are starting to act in the real world, but they can't reliably understand it. Every fact about a place has its own silo: land in GIS maps, power in utility portals, people in census tables, policy in county PDFs. All built for humans, none for agents. Underneath, the data is fragmented across thousands of sources, unstructured and complex. Mireye is building the infrastructure that powers agents to make decisions and act in the real world. One API and MCP to connect agents to the physical world with accurate data enrichment, signals and tools. Our mission is to index every inch of the earth, and make it as queryable as the web.
    geographic-information-system
    location-based
    ai
  • Touchmark
    Touchmark
    Y Combinator LogoS2026
    Active • 2 employees • San Francisco
    Buy tokens over a time window from the provider of your choice at a price fixed today - 10-30% off the on-demand price. If your plans change - resell your reserved capacity to another buyer.
    infrastructure
    ai
  • machine0
    machine0
    Y Combinator LogoS2026
    Active • 1 employees • San Francisco
    machine0 is a CLI for agent compute: `machine0 new mybox` gives your agent a full cloud VM, billed by the minute, from small up to 60 vCPU / 240 GB and 8×H100s. Agents drive it themselves via CLI or MCP. Agent workloads are moving from ephemeral to always-on. A coding agent working on a hard feature runs 6-8 hours. Auto-research agents run for days. machine0 is the environment in which these agents run.
    developer-tools
    cloud-computing
  • Context.dev
    Context.dev
    Y Combinator LogoS2026
    Active • 6 employees • San Francisco
    Context.dev gives AI agents structured data from any website through a single API. We help developers build smarter agents & products that depend on accurate, fresh web data. If you need structured data from the internet, you need Context.dev Proud to power agentic products at Mintlify, Super, Vizzy, Klarna, and 800+ other companies.
    apis
    b2b
    ai
  • Miso Labs
    Miso Labs
    Y Combinator LogoP2026
    Active • 3 employees • San Francisco
    Miso Labs is building the world’s most emotive foundation models for voice. Our goal is to pass the voice Turing test.
  • Twolabs
    Twolabs
    Y Combinator LogoP2026
    Active • 2 employees • San Francisco
    We build verifiable reinforcement-learning environments, evaluations, and proprietary training datasets to improve frontier AI models and robot policies. We focus on complex, underexplored domains where high-quality training data is scarce and existing models fall short. Our long-term goal is to develop self-improving training infrastructure that can autonomously generate increasingly challenging tasks, evaluate performance, and continuously advance model capabilities toward recursive self-improvement.
    robotics
    ai
  • Codag
    Codag
    Y Combinator LogoS2026
    Active • 1 employees • San Francisco
    Codag compresses tool call outputs before agents read them, which reduces token spend by 20-30% while retaining output accuracy. Govern and see analytics on agent dollars across all model providers and internal agents.
    developer-tools
  • RightNow
    RightNow
    Y Combinator LogoF2026
    Active • 2 employees • San Francisco
    RightNow is a GPU kernel lab. Top frontier labs use our optimization tools to get more out of their GPUs, and now we put the same stack behind an API so anyone can own their AI. Run the best open models at full speed per token, or with reserved capacity, dedicated deployments and batch at half price. For enterprises, we optimize your own models down to the kernel
    ai
    infrastructure
    machine-learning
    generative-ai
Loading more companies...