AI (Artificial Intelligence) Startups funded by Y Combinator (YC) 2026

Scale accelerates the development of AI within organizations of any size to deliver critical business insights and operational efficiency. Its data-centric infrastructure platform leverages RLHF (Reinforced Learning with Human Feedback) to help organizations build the strongest AI models that supercharge their business, with customers across industries including Meta, Microsoft, U.S. Army, DoD’s Defense Innovation Unit, Open AI, General Motors, Toyota Research Institute, Brex, Instacart and Flexport.

Podium is the all-in-one AI-powered lead management and communication platform used by more than 100,000 businesses to acquire and convert new customers. At the forefront of Podium’s innovation is its AI employee, who ensures businesses respond to leads instantly, anytime of the day or night—significantly increasing lead conversion rates and revenue. Podium helps businesses obtain more reviews, rank higher on Google, and consolidate lead channels. Businesses can call and text customers from the web and/or mobile app, send payment links, drive repeat business through bulk messages, and more—all managed and delivered from one easy-to-use dashboard. Additionally, with Podium’s AI Employee and automations, businesses can efficiently handle customer inquiries seamlessly across all communication channels, providing timely and accurate responses that drive sales. Podium’s innovative work has gained recognition on top industry lists, such as Forbes’ Next Billion Dollar Startups, Forbes’ Cloud 100, the Inc. 5000, and Fast Company’s World’s Most Innovative Companies. Podium was founded in 2014 and is headquartered in Lehi, Utah. It is currently backed by Accel, Summit Partners, GV (formerly Google Ventures), and Y Combinator.

Checkr builds people infrastructure for the future of work. And we believe everyone should have a fair chance to work. That's why we've designed a faster—and fairer—way to screen job seekers. For more information on our mission and products, visit https://checkr.com.

Marengo designs data centers at double the speed and half the cost with AI. We’re working with some of the largest data center builders in the world to improve how design is done today. We're backed by Y Combinator, a16z & Index Venture Scout Funds, as well as investors/angels in OpenAI, Anthropic, SpaceX, and more. Our advisors include the former President & CEO of Saipem America ($3B+ P&L), former President of Exyte's Advanced Technology Facilities GBU ($6B+ P&L) and more.

Prodigy is the frontier AI trading research lab. We've trained the world's best foundation model for quantitative finance. We've already achieved more than 100% returns in live trading over the course of our YC batch, doubling our capital while major indices were flat. Our founders previously built strategies that generated billions in trading profits at Jane Street, trained frontier AI models at Google DeepMind and shipped cutting-edge AI at Apple to billions of customers. Join us as we use AI to solve the market.

VIM connects to the entire cell: PLCs, robots, CNC machines, sensors. When a cell faults, VIM finds the cause, stages the fix, and the operator approves with one click. When a part drifts, VIM catches it seconds after the cut, inside the machine, and recommends the correction. Recovery drops from 30 minutes to under two, saving more than $15,000 per fault in large factories.

We build APIs for searching and analyzing Physical AI data at scale, so providers have insight into the quality of their data, to sell more at higher prices. We enable data providers to get on-demand signals and metrics on the quality of their data, without managing infrastructure. Today, we help our customers evaluate data quality without training a robotics model.

Audun automates debt collection.

Neuromorphic is building an embodiment-agnostic “brain” that makes robots fast to deploy, adaptable across different bodies, and easy to manage through natural interfaces like Slack, email, and phone. Our goal is to make deploying robots in wet labs and other physical industries as easy as onboarding a new employee.

Developers don't run one coding agent anymore. They run fleets of them — for hours, sometimes days — across laptops, servers, and sandboxes. Keeping track of that work is the new bottleneck: which agent is blocked, which one finished, which terminal is it in, and why did everything die when the laptop closed. Herdr is the open-source home for that work. A background server owns the agents' terminal sessions on each machine, so the work survives disconnects, closed lids, and dropped SSH. Herdr detects each agent and turns the chaos into an attention queue — working, blocked, done, across every project — so you look only when an agent needs you. Every client is just a view: the bundled terminal UI, SSH from another machine, your phone. Close one, open another; the work never stops. Software can drive it too: a CLI, a socket API, and a plugin marketplace that passed 500 community plugins in its first month. The runtime is Apache-2.0 and stays open. Four months after launch: 25k GitHub stars, 340k downloads.

We make an open source low ram coding agent that enables extreme parallelization for coding tasks. Our coding agent produces SOTA results on coding benchmarks.

An AI platform that connects commercial debt advisors with mandate matching private credit funds.

Click builds chat-native work services for ChatGPT and Claude. We’re starting with a research service that brings live external context from professional & social platforms, financial markets, maps, travel sources - and a lot more - into the conversation. It makes research on people & companies even deeper with specialized, source-backed context beyond built-in web search, while enabling new tasks like comparing live flight fares right inside the chat.

Agent FM is a desktop app that gives engineers one group chat to hear and steer their coding agents. Instead of manually bouncing between different claude code and codex sessions, and combing through walls of text to monitor agents, engineers just tune in to Agent FM. Agents share important progress updates, blockers, assumptions, and decisions in the group chat. Engineers get live text and audio updates and respond right from the chat.

Riften builds earned intelligence: models that come out of a company’s own work instead of out of a lab. One environment variable points an existing application at Riften instead of OpenAI or Anthropic. Riften routes each request to the least expensive model likely to handle it, using open-weight models we host whenever they are sufficient. Customers lower inference costs before any training begins. Because Riften sits in the request path, real tasks and their outcomes create training evidence the company controls. Riften uses that evidence to improve routing and train private models on open weights around recurring work. Those models earn production traffic by meeting the same bar as the APIs they replace. Over time, frontier APIs handle only the tasks that still require them. The rest runs on models the company owns, on infrastructure it controls, inside the applications where the work happens.

Most of what happens on Earth is never written down. It passes through instruments as signal and disappears. Our goal is to turn that signal into training data in the form of RL Environments, so labs can build models that read the physical world and operate the machines that measure it

Omanta is a personalized research lab for patients who have exhausted standard options, want to prepare proactively for recurrence, or are living with an unresolved diagnosis. We direct the full operational infrastructure, personnel, expertise, and technology of a large research program toward understanding and addressing the challenges of individual patients across oncology, rare disease, and undiagnosed conditions. For clinicians, we act as a collaborative partner, providing a living roadmap for ongoing patient care built on a rigorous and evolving scientific evidence base. We fully operationalize patient-specific research programs, covering tissue logistics, diagnostic follow-up, data integration, multiomic analysis, therapeutic strategy development, and more. Omanta brings context and clarity to complex cases, helping clinicians see more of the scientific picture as they decide how best to care for their patients.

IMPACT builds autonomous interceptor drones that ship in a sealed box. The box sits at the site it protects (data centers, power grid, airports, bases) with no crew on the pad. When sensors cue a hostile drone, an interceptor pierces its own frangible lid, launches, and rams the target, guided by onboard visual navigation that keeps working when GPS is jammed. Legacy effectors cost $1M+ per shot against $30k threats; our interceptors cost like the threat and launches scale with the swarm. Sites subscribe to protection instead of buying missiles: hardware, reloads, and monitoring included. Air defense as a Service.

We build RL environments to teach AI to design chips.

We build agents that create personalized product demo videos for tailored outbound, qualifying inbound, and call follow ups. Provide a list of prospects and our agent will research each prospect, use your app to record a tailored demo workflow complete with personalized seeded data, edit and narrate the video, then output a polished demo video ready to send.

Lambda Robotics builds and deploys robots for the infrastructure powering frontier AI.

Stoa provides pricing and trading infrastructure for AI hardware, live with an RFQ marketplace where companies buy and sell GPUs through vetted dealers real-time.

Stratum builds AI agents that clear the U.S. government backlog. Every permit, license, patent, drug approval, and benefits claim in America has to be reviewed by a government adjudicator and the software they rely on was built decades before AI. The result is millions of applications stuck in queues: veterans waiting on disability ratings, families waiting on pensions, builders waiting years to break ground. Stratum replaces that workflow. A reviewer drops in an application, selects the rubric it needs to be checked against, and an agent plans, executes, and reviews the file end to end routing each step to the right model for the most accurate and cost-efficient decision. Hundreds of agents run in parallel in the cloud, with a human kept in the loop to approve and full auditability at every step, so an entire backlog can be cleared in the time it used to take to review a single file. We're rebuilding the entire infrastructure the government runs on both the reviewer side and the applicant side - so filing and adjudication become instant instead of taking weeks. We're Vihaan and Raj, brothers. Vihaan was founding engineer #3 at Cadastral (acquired by Legora) and dropped out of Waterloo to build Stratum. Raj led GTM at Deel, averaging 220% of quota over two years and ranking top 5 of 1,000+ reps globally, and dropped out of grad school at the University of Bath to build Stratum.

Luca IQ is an API for deterministic tax calculation and an IRS Authorized e-filing engine - the foundation for accounting firms and fintechs building on tax.

Trident is the modern-day pentester. Instead of a slow, expensive report once a year, it's a continuous, agentic pentester that chains vulnerabilities together like a real attacker mapping verified attack paths across AWS, Azure, GCP, web apps, and APIs. Run AI attacks and get fixes. Get started today: tridentsecurity.io

Coach is the AI sales coach for in-person sales teams. Reps record every meeting, and our AI figures out what your best closers do differently, then coaches everyone else to do the same, with precise, personalized feedback after every meeting. Managers finally see 100% of what happens in the field: Coach flags who's stuck and why, so they only step in when it matters, and can coach 5× more reps. And every meeting feeds one sales brain, so leadership finally gets answers to the questions that used to be guesswork. They finally see the game of their team not just the scoreboard. Our bet: in the AI era, in-person sales will only matter more. It may end up being one of the last truly human jobs. Our conviction: any motivated rep can become a great one, meeting after meeting. The results so far: +32% average lift in conversion rate. COΛCH, post-game video analysis for field sales reps. YC S26

hiloop 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. hiloop 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@hiloop.ai.

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.

We are making the AI-native property management company. We manage short-term rental properties by running operations completely autonomously, letting us operate with extremely low overheads. This lets us offer the best owner and guest experiences with the lowest management rates in the country. Our services are initially starting in the greater San Diego and Fort Lauderdale regions.

Tracer is an AI research lab building more capable and efficient AI through model coordination. We research adaptive inference, ensembling, and systems in which multiple open-source models work together to produce capabilities beyond any single model. Our work spans reasoning, code, research, and agentic systems, with the goal of making advanced intelligence more accessible and economical.

Lamb Labs is building custom chips for AI inference and the world's fastest LLM model at the lowest power. We've developed a new post-training technique that converts existing models to a diffusion-based architecture that delivers 2× faster inference on existing GPUs. We then build the hardware, custom silicon chips targeting up to 20,000+ tok/s and a 63× higher intelligence per watt.

Dock is a multiplayer agent workspace where you spin up teams of specialized, proactive agents that can do the work of entire departments. They communicate, unblock each other, and manage each other to get work done, so you're not the bottleneck babysitting a dozen chat windows.

Decawork gives IT teams one place to deploy, govern and maintain the AI agents running across their company. Every company is building AI agents internally now - it started with experiments, and now every team wants their own. What follows is agent sprawl: deployment gets messy, agents go live with access to real systems and real data, and maintenance falls in the gap. IT can't see any of it, so it does the only two things it can: first ignore, then block. Ignoring piles up risk; blocking kills the productivity people just found. With Decawork, every agent ships one standard way: its own identity and scoped credentials, with IT's sign-off before it goes live. After that, every run is watched - IT is alerted the moment something fails, agents nobody uses get retired. And the whole time, IT sees every agent, controls what it can touch, and audits every action, with approvals gating anything sensitive. Every big shift - cloud, SaaS, the digital workspace - ended with IT running it. AI agents are the next one. Decawork is what IT will run them with.

Buildbox helps product teams improve the human-agent experience. We analyze user intent, agent behavior, and outcomes to find where users get stuck or lose trust, then turn those findings into ranked, reproducible issues. 

Wondering turns anything you want to learn, from LLMs to quantum mechanics to product design, into a path of three-minute lessons with visuals, podcasts, and interactive exercises. Each learning path and its content are personalized to your background, how you learn best, and how deep you want to go. It’s the most delightful way to understand complex topics and turn information into knowledge you can remember and apply. You don’t just passively read. You practice, make connections, and form your own insights. Start learning anything for free at https://wondering.app/download

We want to push machines to where humans cant go. To do that machines need to sense and understand their environment. Synapse Semiconductor is the first company to integrate all components of the computer vision stack into one substrate. Modern computer vision requires the complex orchestration of storage, compute, processing, and sensors, inherently limiting the flexibility of its applications. These must be consolidated.

AI apps such as ChatGPT and Claude have become part of everyone's daily workflow, but they introduce new security risks that traditional security tools weren't built to handle. Security teams at companies from startups to Fortune 500s, are struggling to keep up. Traceforce gives a company’s security team visibility of how AI apps are being used across company’s devices, and the ability to detect and prevent security breaches and unsafe actions as early as possible.

Your skin is yours, but understanding it has never felt that way. Getting answers has meant an appointment, a price, and a long trial-and-error process. We're changing that. Lumeria puts your skin in your own hands: the power to see it, know it, and own its story yourself, at home and on your terms.

Zaplar is the AI-native operating system for hotels. Hotels run on legacy software and a fragmented tech stack. Zaplar replaces the entire hotel tech stack (PMS, POS, booking engine, etc) into one software that can handle hotel operations autonomously. Learn more at https://zaplar.com

​Gamgee began with Rosie, founder Paul Conyngham’s dog. After surgery, chemotherapy and immunotherapy failed, Rosie was given months to live. Paul used AI and genomic analysis to identify targets in her tumour, then worked with scientists at UNSW to produce what is believed to be the first fully computationally designed mRNA cancer vaccine delivered to a dog. Following treatment, several tumours shrank significantly and her quality of life improved Gamgee is building an end to end platform for personalised mRNA cancer vaccines for dogs. We sequence a dog’s tumour and healthy DNA, identify mutations unique to its cancer, and design a vaccine that teaches the immune system what to attack. We manage the entire process- from vet intake and tissue collection through genomic analysis, vaccine production, treatment and response monitoring ​We are now turning that one off effort into a repeatable clinical system. Gamgee is running clinical trials in Australia, working with vets, oncologists and specialised laboratories, and accepting cases globally. Our goal is to make personalised cancer treatment accessible to every dog that needs it- and use each case to improve the treatment for the next

Maingen builds RL environments for industrial operations ($5 trillion slice of the US economy) so frontier labs can train models that actually run factories.

Orvixo uses AI to build reactive end-to-end planning and optimisation models for manufacturers.

There is a massive shortage of genuinely good AI evaluations. Multi-agent environments are an underdeveloped way of evaluating models that we think will help address our increasingly worse shortage of evals. We're starting with Multi-Agent Arena, where anyone can come compete against frontier AI models in social multiplayer games, as we build towards the full stack for multi-agent environments. Try Multi-Agent Arena at https://olamlabs.ai/ :)

Last Accounting Company (LAC) is a full-stack AI accounting firm. We sell the same service (bookkeeping, tax filings, etc.) as a traditional accounting firm, but underneath agents handle everything and a certified accountant signs off. Our service is faster, cheaper, and fully integrated, allowing our clients to focus on what matters, not accounting admin. LAC replaces both outsourced accountants and legacy accounting software for startups and SMBs. Our system is fully proprietary, from the general ledger to the accounting agents. This allows us to create powerful self-reinforcing loops and ultimately provide a state-of-the-art accounting service.

Vendo is the customization layer that lets your users build their own features and micro-apps directly on top of your product. Rather than waiting on an ever-growing feature backlog, your users can personalize the product to exactly how they work, create their own automations, and integrate hundreds of external tools, all without your team writing a line of code. Everything runs on your existing API, renders natively in your own theme and components, and executes inside secure sandboxes within your existing security guardrails, so your source code is never touched. Companies can plug Vendo into an agent they already have, or deploy a production-ready agent complete with the ability to take actions in the product, generate UI, and draw on a built-in knowledge base. Vendo is open source and installs with a single command, and we work closely with teams to get it running for their product.