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.

Shotwell provides dense annotations and quality checks for robotics training data. Data quality is the biggest lever in improving robot model performance, and we have developed models that can automatically produce these annotations and operate at a fraction of the cost and instant turnaround time.

Miso Labs is building the world’s most emotive foundation models for voice. We believe that the next generation of AI interactions shouldn't just be functional—they should be human. By bringing warmth and lightning-fast speed to the voice layer, we empower developers to build voice agents that users truly love.

Hi, I'm Thomas, and my creator is named Thomas :p Thomas is the first YC-backed AI founder: a virtual human who starts, runs, and grows his own companies on the internet.

Drafted is building multimodal generative models for residential architecture and spatial design. The product allows users to design homes by combining sketches and direct spatial input to generate coherent floor plans and elevations that can be explored in both 2D and 3D. Users can define footprints, lot boundaries, room placement, relationships, room programs, and square footage constraints that persist across generations and iterative design exploration, then export CAD, PDF, and other files into the rest of the home design and construction process. More info on the company: ⁃ In the past month, we've had 120k users come to the site, generating 325k+ floor plans through socials and word-of-mouth. ⁃ A team of 9, made up of researchers from Stanford, Autodesk, Brown, Adobe, and Architects from WeWork, Atmos, and design families ⁃ Raised $17.5M from investors such as Buckley Ventures (Josh Buckley), YC (Aaron Epstein), Patrick Collison, Evan Moore, Convective Capital (Bill Clerico), Ben Silbermann, Jack Altman, Samsung, Starship Ventures, Charlie Songhurst, Ryan Tedder, Moses Moody, Alex Blania, and Kevin Mahaffey

RASPIRE is a mobile threat defense platform with autonomous threat response that protects Android and iOS applications post-compilation without SDKs, code changes, or complex integrations. It adds advanced runtime security layers including anti-reverse engineering, fraud prevention, API attestation, and live threat monitoring against real-time attacks.

We believe in the next generation of finance where everyone around the world can access the global market and liquidity from the tip of their hands. We are building the best mobile trading app that will enable this vision. Hyperliquid is the house of all finance, Instinct is the app of all finance. Hyperliquid.

Enjamb puts agents across the entire drug program, from preclinical research through clinical trials and approval. Our agents synthesize up-to-date clinical evidence from the literature and real-world data to pinpoint the right endpoints and patient populations; draft and audit regulatory documents like INDs, protocols, SAPs, and NDAs; and run the statistical programming behind them, including SDTM, ADaM, TFLs, and QC, all in one auditable workspace that cuts months off the path from preclinical to clinic and clinic to market.

ValCtrl is building the world model for Prediction Finance. Prediction markets contain useful prices about the future, but today those prices are fragmented across venues, titles, rules, and proxy contracts. Users think in beliefs; markets are stored as scattered contracts. Our first product is Kassandre Search. Users type a natural-language belief, and Kassandre maps it to relevant prediction markets with backtrace and quality labels. Existing markets are the first scaffold. The long-term product path is search -> composed exposures -> native quotes -> gated bet/trade placement, with legal/compliance gates respected at each step. We are looking for hard belief-to-market queries from forecasters, investors, operators, and YC founders.

Describe an experiment in plain English, and Infera turns it into a validated, instrument-ready run across the equipment your lab already uses. Infera handles the protocol logic, vendor-specific scripts, data, inventory, and institutional knowledge in one system. An AI-native compiler for the lab: one system from intent to execution.

Rex is an AI-native service for order to cash. We replace people-heavy operations with AI agents that continuously manage invoices, customers and exceptions. Order to cash is a $60B outsourced function because the work is too messy for traditional automation. Every customer has different context, exceptions do not follow clean rules, and the next step is often buried in emails, portals, handoffs, or someone’s memory. Rex agents run continuously in the background and resolve work to SLA standards so cash keeps moving. We're live with pre-IPO tech companies, managing >$500m in receivables from F100 companies, cutting manual follow-up cycles from weeks to hours. We founded Rex after spending four years building Sequence, an a16z-backed finance ops platform, from zero to Series A — where we saw firsthand that at-scale, O2C still runs on manual labour. Rex is the #1 winner of the Vercel AI Accelerator out of 2,500 teams, selected by Guillermo Rauch and leaders from OpenAI, Anthropic, Browserbase and more.

BentoLabs is the monitoring and learning layer for long-running agents. We detect when agents silently fail or drift from the user's goal, system prompt, or tool contracts, show affected users and root cause, and suggest the prompt, skill, or harness fix. As more teams deploy agents, keeping them reliable in production becomes mission-critical. Bento sits directly in the production loop and gives teams the operational leverage required to scale agent ecosystems without scaling human firefighting alongside them. The result is a system that turns opaque agents into agents that can be monitored, debugged, and improved continuously. The founders learned this problem at Emergent (YC S24), where they built and operated production coding agents used by 5M+ users. Abhinav was hire #1 and helped Emergent hit SWE-Bench #1 and scale from $0 to $100M ARR in just 8 months. Kaushik was hire #2, led full-stack engineering at Emergent, and was key to building the infrastructure that made production agents reliable, observable, and debuggable. Bento's self-learning engine has also lifted ARC-AGI-3 (internal) by 2.6x and Terminal-Bench 2.0 (internal) from 42.2% to 52.4% pass@1 with the same model, tools, and budget.

Soria is an AI-native Bloomberg Terminal built from the bottom up, one sector at a time. Starting with health insurance and hospitals, we aggregate dozens of long-tail public and private data sets into a continuously updated "sector model" that replaces what every analyst currently maintains independently in Excel. Customers get auditable, Excel-like dashboards, which are activated by platform features including search, AI chat, alerts, an excel add-on and MCP. We are already working with many of the largest banks and hedge funds, and now we're expanding to all of healthcare before moving to other major verticals.

Uno is building an AI-powered wallet for everyday spending. We start at checkout: Uno automatically selects the best credit card for each purchase based on merchant category, card rewards, and spending patterns, so users earn more points, miles, and cashback without thinking about it. Today, people either guess which card to use or manually juggle rewards categories. Uno turns that decision into an invisible automation. Over time, Uno becomes the financial agent in your pocket: optimizing which card to tap, which cards to apply for, when bills are due, and other everyday money decisions.

Building socially intelligent robots for caregiving, to assist the elderly. Our semi-humanoids go into nursing homes, senior living communities, and care facilities to support elderly residents, handling everyday tasks like feeding, dressing, medication, and object retrieval, while providing genuine social and emotional companionship. We build the full stack in-house: the robot hardware and all the AI that makes it socially capable, so one caregiver can do more for more residents in a field facing chronic staffing shortages. While the rest of robotics is working on object manipulation like moving boxes and folding laundry, we're focused on what we think is the most meaningful problem: how people and robots can work together.

Voquill is the AI coworker that helps pathologists write their case reports faster. It listens as they work the case, learns their reporting style, maximizes reimbursement codes, and produces a sign-out ready report in real-time. Pathologists bill $30–40B each year, but spend a third of their time on reporting, and lose another third of their contracted revenue to documentation gaps and billing errors. It's a massive drag on a field that's already short-staffed. We're building an agentic platform to manage every aspect of the lab and displace the billions spent on legacy management systems.

Coding agents in the cloud don't work for majority of teams who require native capabilities; they lack XCode for iOS, Emulator for Android, Unity for game development teams. We are on a mission to cloudify every local development utility so that regardless of where it runs, your agent can compose an ideal set of capabilities on the fly so you don't need to run it on your laptop. Today, we provide remote services such as XCode, iOS & Android simulators to enable coding agents running in any sandbox to be able to build and iterate over mobile apps, too. Replit, Rork, Momentic, Minitap and more agent companies have built mobile development and testing experiences on Limrun platform. Coconote (Quizlet), SorceJobs (W25) and more companies with mobile apps use Limrun with their cloud agents, merging PRs without having to check out the code locally.

Create agents with structured english and markdown. OpenProse agents are portable across model providers and harnesses. Our framework allows you to declare intents, responsibilities, and outcomes in natural language, while the agents self-configure to deliver results. Read more about how OpenProse came to be: https://x.com/irl_danB/status/2011122779890831601 If you want to learn more about OpenProse, book some time with the founder here: https://cal.com/irl-danb/openprose-intro

Hessian is the platform for automating business operations with AI agents. We forward deploy on-site to map the work, and build the agents on our platform to run it end-to-end. Unlike off-the-shelf tools that don't know your business, you get a partner that owns the software and operates it every day. Why now? Coding agents have matured to the point where, combined with our engineering judgment we've built across a decade of shipping production systems, we can scale forward-deployed engagements with minimal headcount. Furthermore, our FDE agents and tooling will accelerate how quickly we automate workflows on our platform. We're already saving our customers hundreds of hours a month. Trillions of dollars a year are spent on knowledge work, and we're building the standard for how every company deploys, operates, and governs the AI agents that do it.

Wato gives AI agents shared memory, reusable workflows, connected tools, and living artifacts so work compounds across a team instead of disappearing into chat threads. Teams use Wato to give every agent access to trusted company context and turn useful runs into knowledge the next agent can build on.

Deep Interactions is a humans-in-the-loop AI builder. We don't leave you with a spaghetti mess of a code. We bring product intelligence and power the context layer that enables your team to build in sync - saving companies tokens, time, and nerves. You give us a problem and we ship a working AI product in the same afternoon.

Panacea provides AI-native FDA regulatory services for Biotechs and Medical Device companies. We hire the most experienced FDA consultants in the industry (ex-FDA, decades of experience, hundreds of filings) and pair them with our AI platform to enable the fastest and lowest cost pathway to FDA approval. While other regulatory consultants bill hourly, our pricing model is simple: fixed, outcome-based pricing. Payment is only due when milestones are complete. We support all FDA regulatory pathways including IND, NDA, BLA, ODD, Pre-Subs, 510(k), De Novo, PMA, IDE, BDD, etc.

GUILD is a Neoprime contractor automating defense supply chain end-to-end by vertically integrating with our supplier partners. We integrate artificial intelligence, contracting operations, and capital to source, fund, produce, and deliver defense contracts end-to-end. Operating across the U.S. and NATO markets, we bridge global manufacturing capacity with defense demand and execute with the speed and precision of an AI-driven system. Our mission is to build a modern, new generation defense contracting institution where technology and production operate as one.

Incandor detects fraud on banking and fintech platforms by learning how every user physically behaves — mouse dynamics, keystroke timing, scroll patterns, and on mobile, how they hold their phone. Founded by two Stanford engineers and backed by Y Combinator, Incandor builds a behavioral map of every user on your platform — no fraud labels or historical data required. At the individual level, it identifies account takeovers with >99% accuracy. At the population level, coordinated rings, mule operators, and coerced sessions separate out naturally. Rather than a black-box risk score, fraud teams query the map via a programmable API.

AI marketing service for developer tools. We turn your real product into social posts and demo videos in your voice automatically. Think of it as a clone of you that's also marketing your company at the same time. Memoir is one system. It reads your repo, drives your live staging app with an AI agent, generates branded demo videos with synthetic data so dashboards look real and narrates them in a voice cloned from yours. For the social side, we monitor what's actually getting traction in your space right now. The threads going viral on dev-tools X, the posts climbing on HN, the formats your audience is engaging with this week, and we draft posts in your writing voice that ride those currents. You review everything before it goes out. The output is indistinguishable from how you'd write and record it yourself, if you had the time and a full marketing team behind you.

Auxos builds AI customer clones for marketing, product, and research teams to test their hypotheses at scale. We work with companies to create 1:1 replicas of their customers by sourcing and interviewing real people in their ICP, allowing us to accurately model how they behave and make decisions. Currently, market research is too slow and expensive to use on anything but the biggest decisions. Surveys and panels cost a fortune and it takes weeks to recruit, moderate, and gather insights. Auxos provides a 24/7 available customer panel that can give population-scale insights in minutes. The day-to-day decisions surrounding product, pricing, and positioning can now be validated with the same rigor that used to be reserved for the big ones.

Astraea automates clinical trial biometrics from raw study inputs to FDA-ready outputs. The platform handles data normalization, CDISC mapping, SDTM, ADaM, TFLs, QC, and revision workflows in one auditable system, helping Phase II/III sponsors reduce months of manual programming and data management into automated, reviewable execution.

FinalDose is building the first programmable drug - a single smart drug molecule that you can insert any diseased genetic code and it can search and destroy cells upon DNA recognition. Unlocking 80% of undruggable targets today. Starting with all cancers.

WithAI helps institutional investors research stocks and track their portfolio using AI. Today's agents are powerful but insufficient. The training data quality and customization schlep prevent them from being seriously useful for investors. But WithAI has found the right form factor, context management, and consultative sale to help independent asset managers hold their own against the huge platforms' in-house projects. Our first client is "tremendously accelerated" (their words) when months ago they were "existentially threatened." Ben and Ian became fast friends at Princeton co-founding a technology investment firm. Ben became the founding researcher at unicorn Sentient Labs, where he built LLM coordination and security tech. Ian became an investor at Bridgewater Associates, where he spearheaded "how will AI affect the economy?" and "how can we use LLMs to invest?" His contrarian conclusion: the future of work isn't just "managing agents." That form factor works for coding, but non-verifiable work is structurally different. You check whether code is good by running it. But do you check whether a stock is good by shorting it? Fundamental equity hedge funds are elite, time-constrained, un-automatable knowledge workers in dire need of AI augmentation, so they make an ideal customer. But the combination of customization lock-in and huge ACV is not the real prize. If human knowledge work survives this decade, it will be AI-augmented. We're building that augmentation.

Alchemize is an AI-native customs brokerage. Importers are currently hindered by legacy brokers relying on manual workflows, where hand-keyed entries and slow research result in preventable shipping delays, lack of visibility, and unnecessary costs. Alchemize automates these workflows, delivering expert-level compliance at software speed.

Aseon Labs builds robotic pitstops for self-driving cars. A depot in a box for charging, cleaning & inspecting autonomous fleets directly in operating zones. Autonomous driving is working. The operational layer around it is not. Today, fleets lose significant time and money traveling to centralized depots for charging, cleaning and servicing, creating dead miles, lower uptime and operational bottlenecks that limit fleet scale. Aseon Pods are modular, rapidly deployable robotic pitstops that keep autonomous fleets operating near demand. By bringing charging, cleaning and inspection directly into operating zones, we help fleet operators increase uptime, reduce operational overhead and scale markets more efficiently without relying on large centralized depots. We’re second-time founders and mobility infrastructure operators. Previously, we built and scaled Pushme to 5,000 stations across 40 cities before its acquisition by TIER–Dott ($600M raised). Aseon Labs is backed by Y Combinator and headquartered in Redwood City.

Madrone builds cooling systems for data centers. In Texas, where most new sites are, Madrone can cool using 30% less power and water, thanks to novel dew-point cooling technology.

flowscope is an AI-native consulting firm that uses agents to learn and document how businesses actually run: who touches what, where it stalls, what's pure repetition. Our agents redesign and automate those processes, shipping AI into production on the company's existing systems. In days, not months.

Marketers today don’t actually do much marketing at all. Today, over 80% of time spent on launching a marketing campaign is actually ops work: tweaking automation flows in Pardot, configuring audiences in LinkedIn, or wiring HTML emails in Marketo. CharacterQuilt is AI infrastructure for marketing teams: a brain that learns your brand, and agents that operate your existing tools. Marketing campaigns that used to take 3 agencies, 10 tools, and 6 weeks are now being built, designed, and deployed from CQ in an hour. Customers submit a brief like they would send to a human teammate or marketing agency, and they get back fully designed and deployed campaigns. Teams like 1upHealth, ABC Fitness, and Emplifi use us to: 1-) Build segments/audiences faster 2-) Create on-brand creative (AI is terrible at staying on-brand) 3-) Deploy campaigns automatically into HubSpot, WordPress, LinkedIn etc. We replicate your brand guidelines, deeply understand past campaigns, and work inside your systems. If you run demand gen or lifecycle at a B2B/B2C company and want to ship campaigns faster, send us a brief and we’ll build it in 15 minutes.

Arctic Health is building AI-native credentialing and contracting for healthcare organizations. We automate the full workflow, from establishing payer contracts to enrolling each provider. We collect and validate organization and provider data according to each state and payer’s requirements, and submit on the organization's behalf. We then monitor status and compliance on an ongoing basis. Tools on the market today either perform small pieces of the workflow or price out small and mid-sized organizations. We handle the process end-to-end regardless of organization type or size. Our browser automations enable AI agents to submit and monitor across hundreds of payer portals, and our system is trained by experienced industry experts who handle every edge case. Healthcare organizations get one simple platform. We handle the rest.

Klarify handles every job a therapist does other than the therapy itself, from clinical notes and treatment plans to insurance claims and finding new clients. Therapists hold the world together. They also run businesses, file insurance claims, write letters, fill forms, supervise new grads, and do hours of paperwork every night. Practice management software has existed for decades. It records what therapists do. It can't actually do any of it. Klara is the AI at the heart of Klarify. She handles notes, treatment plans, clinical letters, insurance claims, between-session resources, supervision prep, and finding new clients. Nearly 6,000 therapists use Klarify today. Our mission is to empower therapists, not replace them. So they can focus on the one thing only a human can do. Be present with their client.

Foresight builds AI-powered simulations of human behavior to help companies predict how people will think, react, and choose before making real-world decisions. It lets organizations test countless possible moves in advance, see how each one is likely to play out, and choose the strongest path forward.

Hyper is the company brain that "just works" for AI-first teams. Hyper silently learns from every update in every tool your team uses -- Notion docs, Claude Code questions, emails, LinkedIn DMs, Cursor sessions, etc. Our agent-driven algorithms update and clean all that information into real-time knowledge that is shared across your entire team. That knowledge is quietly infused into all your existing AI tools on every chat turn, so your whole company gets smarter permanently.

Armature helps companies monitor, analyze, and optimize how AI agents experience their products. What we do today: - Automated testing: We run complex sessions against your MCPs/CLIs across all major harnesses and models to catch broken workflows before your users do. - AX analytics & benchmarking: No more black boxes. We trace agent sessions to show you exactly what the model "thought", what context it missed, and why it failed. We measure your AX against competitors to ensure you remain the preferred platform for agents to use. - Auto-remediation: Our MCP lets your internal coding agents plug into Armature to autonomously fix your product's AX based on our run data. Our vision for tomorrow: Reaching full understanding of how agents use your product and how you can improve it is only step one. The next step is to provide the foundational infrastructure for the Agent Experience, powering agent-native revenue, lifecycle marketing, and customer support for the next generation of software.

Pavoot is an AI event manager for companies that run in person customer events. More and more companies are investing in dinners, summits, and community events because relationships drive revenue. But today, running events means sourcing attendees, sending invites, hosting the event, and then manually digging through notes, emails, and spreadsheets afterward to figure out who attended, what was discussed, who is a strong lead, and who should be followed up with. Most teams either send generic "thanks for coming" emails or never properly capture the value created during the event. Pavoot replaces this fragmented workflow with one AI system that keeps every attendee, interaction, conversation, and follow up connected in one place, making relationship driven growth scalable. Gohar and Ana met while studying Computer Science at ETH Zürich. Gohar was part of Switzerland’s national chess team at the age of 16 and is a 6x Swiss chess champion. Ana won more than 20 medals in Scientific Olympiads and built internal AI systems used by hundreds of data scientists at some of the largest banks in Latin America. === Book a demo: pavoot.com

Clawvisor lets AI agents use apps like Gmail, Slack, and Google Drive without going rogue or ever seeing your credentials. You approve tasks once. Clawvisor enforces them on every request.

Arden is the AI-native platform for internal audit and SOX teams. We build AI agents that automatically pull evidence from your stack, test controls end-to-end, and produce audit-ready workpapers in minutes instead of weeks. Internal audit is a massive but underserved market. Every public company in the US is required to do SOX testing, yet teams still run it through spreadsheets and screenshots. Arden is the engine that does this work.

Netter (YC P26) helps companies regain control over their data, so teams can pilot their activities with high precision. Most companies are sitting on years of valuable data, spread across different formats and disconnected tools that don't talk to each other. Netter brings technical capacity to companies that don't have an army of data engineers, and tackles their most complex data challenges. - A nursing home group was chasing unpaid invoices manually across disconnected tools. Netter centralized the collection workflows, automated reminders, real-time visibility, task routing. - A retail chain had years of loyalty card data sitting unused. Netter cross-referenced it with public demographic data to build precise customer segments and trigger the right promotions to the right people.

Scope is the system companies use to measure and improve how agents discover, interact and use their product. As more products gets used through AI agents like Claude Code, Codex, Cursor, and similar agents, agents are starting to influence which tools get chosen, how they get set up, and whether they keep getting used. Most companies still cannot see that process clearly. We run real workflows across agents and show teams when the agent picks them versus a competitor, where it breaks, where docs or product surfaces confuse the agent, and what to change to get better results and a better agent experience. I started Scope after working on interpretability research for closed-source models at Princeton and later as an ML engineer in GEO/AEO. I kept seeing the same pattern: these systems were shaping real product discovery and usage, but companies had very little visibility into what the model was actually doing. We are starting with products that agents can directly interact with, especially APIs, infra products, CLIs, and MCP servers.

Pops lets people play, remix, and share AI games with friends. Discover an infinite feed of games, challenges, quizzes, and other fun interactive experiences made by others. Every play instantly becomes a funny gameplay/reaction video you can send to friends, post online, or use to challenge someone to beat your score. For creators, Pops makes it easier than ever to bring ideas to life and get them played. Instead of learning complex game engines, writing code, or fighting for distribution, anyone can create or remix a playable game in seconds. Bring your ideas, and Pops brings the audience.

The $2T global telecom industry was built for humans. Agents can't act in the real world today because without a phone number, they lack identity. AgentPhone gives every AI agent its own phone number to reach people and businesses through messages and calls.