AI (Artificial Intelligence) Startups funded by Y Combinator (YC) in San Francisco 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.

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.

An open-source toolkit to capture traces, evaluate cheaper models against benchmarks, and ship specialist routes you own. Capture traces from LLM production workflows with a single install that deploys within coding agents you already use. Hosted infrastructure is optional. Evaluate the captured traces and set a benchmark for success. Every future model switch meets or exceeds it in A/B testing. Train and fine-tune a new model on prompts and weights you always own. Start locally and scale into cloud-hosted processes as you see success. Deploy a new model only when the held-out eval is beaten. Serve it wherever you want. Production data feeds back into training and compounds performance over time. Test. Learn. Deploy. Repeat.

An AI-native media infrastructure company that provides natural language video editing systems and vision-driven clip orchestration engines. Our platform allows enterprise teams to drop in raw footage or corporate documents and execute precise timeline modifications, clip stitching, and pixel-perfect graphic overlays entirely through conversational interfaces.

Alkera is a data engineering and data science agent that works in your IDE or CLI. If you’re tired of agents taking forever just to produce false, or even dangerous, reports about your data, then Alkera is for you. Our agent is equipped with your data’s history and lineage, your team’s knowledge, and tailored plugins for data apps such as Snowflake and dbt. Simply install and connect your services to begin doing trusted agentic data work. We also offer VPC & on-prem deployment options for when data can't leave your boundary.

Skymerse is building one AI system for airline flight operations: from planning and monitoring in the operations center to real-time support in the cockpit. It is developing specialized aeronautical models that understand the data, constraints, and procedures behind each flight, replacing fragmented tools and manual interpretation. Notamify, its first product, already serves airline customers and thousands pilots and aviation professionals.

Zomma builds Computer Use Agents that do the repetitive screen work in finance: logging into portals, pulling reports, reconciling books, and filling forms across systems that will never get an API. Our agents use the same screens your ops team uses today, so there is no integration project. Every action stops for human approval before anything submits. Hedge funds, asset managers, and fintech firms use Zomma for reconciliation, client onboarding, KYC, and investor reporting: the work that was never automatable until now.

Torus lets engineering firms and project owners build infrastructure 10x faster to reindustrialize America and the West. We're working with multiple Fortune 500 companies to build critical infrastructure including data centers and critical energy installations.

Trope is the AI-native ERP implementation platform that helps partners deliver successful projects faster. Trope understands the entire project lifecycle from pre-sales to go-live and beyond. Agents take action across documents, configurations, and code. For implementation partners, it’s like having a project manager, consultant, and developer working 24/7.

screenpipe record and learns how you work and turns it into a searchable memory, SOPs, and AI agents. Open source, local.

TrustAI understands your context across the browser (CRMs, internal tools, Slack, GitHub) and suggests personalized automations you can run in one click. Accept, tweak, or ignore, then keep working while we handle the busywork in the background.

Robocurve builds open-source tools and independent benchmarks to measure how well robots can do real-world jobs. Instead of relying on unverified demo videos from frontier labs, we score their models on reproducible benchmarks that anyone can trust. Today there are no well-run, standardized robotics benchmarks. Labs evaluate in-house, and no independent group has stepped in to run continuous benchmarking as a service. The result is that no one actually knows how good anyone else is, or where the real frontier sits. Good benchmarks require operating and maintaining physical hardware and real-world setups. Simulation only goes so far, since models that look strong in sim can show large performance gaps once deployed in the real world. Building real-world benchmarks means coordinating job-domain experts, evals engineering, and hands-on robotics all at once. Frontier labs are targeting general-purpose robotics by 2028, yet the field of robotics evals barely exists. Whoever builds the trusted measure of robot capability becomes the reference everyone relies on. We combine backgrounds in AI evals and robotics to build and grow this field as fast as possible. We already shipped v1 of our open-source framework, Inspect Robots, and ran our first pilot scoring a frontier model on a real robot.

LemonLime enables any business to deploy custom AI automations in minutes. It connects to your existing tools to pull information across sources, studies your business, and instantly creates specialized agents designed to support your team across departments. Don't know where to start? LemonLime automatically surfaces suggested automations that can be implemented with a single click.

Justinian is an applied AI company changing the way companies and governments work together. Our first focus is building Justinian Government Affairs, an AI-native government affairs firm helping early-stage companies monitor regulations, lobby policymakers, and win government contracts.

Risklytics builds AI risk models that score natural disaster risk property by property, so insurers can write the good risks hiding inside 'high-risk' zones instead of declining entire ZIP codes. One API call returns the risk score, the drivers behind it, and the cheapest retrofit path to insurability.

We are building the identity and communication layer for AI agents: giving agents their own email, phone, iMessage, and internet address. Every interaction flows through Inkbox, so agents are enhanced with multi-channel context.

Instance is ground truth for robot learning. Labs generate thousands of robot episodes a week — far more than anyone can watch — and the mislabels nobody catches poison training data, evals, and RL rewards. Give us a task description and video rollouts, and we return a success verdict with evidence. Teams use Instance to clean training data, run evals they can trust, and get reward signals for reinforcement learning without hand-labeling. We benchmark our verifier against human-labeled episodes across 8 benchmarks and 7 robot platforms, where it outperforms frontier VLMs in accuracy and latency. We're MIT computer scientists & friends since middle school, with backgrounds at NASA JPL, SpaceX, and AWS.

Definite builds AI agents for the financial back office. Before any output touches your systems, our verification engine checks it against the source documents. Numbers get recomputed from contracts, rate tables, and ledger records, and wrong outputs go back to the agent with the reason, like compiler errors. They don't ship. We met our first year studying computer science at the University of Waterloo, and interned at Meta, Optiver, and BitGo, places where one unverified number moving through a system has real consequences. Talking to teams deploying agents, we kept seeing the same failure: back-office pilots that demo well but never graduate, because nobody trusts the outputs enough to stop re-checking them. Today teams check agent work with deterministic rules, an LLM judge, and human sampling. The middle is the gap. Judge models miss what the agent missed, rules only cover what someone wrote down, so a human reviews everything and the economics never improve. This is hard because verification has to be independent of the model doing the work. Recomputing the output has to be done from source documents instead of asking a model to grade its own answer. Every back office is getting agents. Whoever makes agent output trustworthy enough to run without a human behind it owns the layer everything else depends on. The same way finance built approval controls before money moves, verified execution becomes mandatory infrastructure. We build it, and the agents that run on it.

An applied AI research lab building world models for agents. We simulate reality faster and more accurately based on the technology self-driving uses to get 99.7% reconstruction fidelity.

Perceptron ML builds the system that turns real-world signals into action. Today we help law firms discover new legal matters the moment they break and launch the campaigns to go after them in hours, all under human approval. The same engine that detects a signal and acts on it fast can serve any business that wins by moving first.

Denta is the practice management platform that puts dental practices on autopilot. When a dentist starts a new practice, instead of finding an accounting firm, hiring two front desk workers, and then spending their nights posting insurance payments, they use Denta. I started Denta to help operate my Dad's 10 practices in Iowa. In supporting those, I realized the most painful part was both people and agents having messy access to fragmented data. So I solved for that and now spend my time automating work for dentists!

Florin brings banking, payments, cards, and treasury into one account built for startups - so founders can hold, move, and grow their capital without stitching together half a dozen tools. Unlike platforms that resell another bank's rails, we're building the infrastructure ourselves, end to end, so money moves in real time on a single ledger we own. We start with an invite-only membership for venture-backed companies and expand from there. Named after the florin, the coin that anchored global trade for three centuries, with infrastructure built for how money moves now.

Poth is an agentic research platform that connects to your company data, creates hypotheses about what might be causing problems, and validates them using both existing data and adaptive user feedback.

Insurf is the AI-native data and decision layer for health insurance. Our first product, Inveto, automatically turns denied healthcare claims into source-cited, physician-attested appeals, recovering revenue practices lose today. Every claim it resolves builds the coverage-decision graph, the proprietary dataset of real payer behavior that powers our next product, Surely, which prices the true cost of a health plan for employers and brokers.

We write the algorithms to turn brain cells into compute.

The best investments come from conviction, and conviction comes from understanding a market deeply: its customers, its competitors, and where it's heading. Today that understanding is slow and expensive to build. Reaching the right people takes weeks, a commercial study costs $300k+, and each one starts from scratch. LATO does this research end-to-end, combining public data and the fund's own knowledge with first-hand interviews of a market's customers, experts, and operators. This research grounds a simulation of the market to predict how market participants behave.

Bloomy helps students learn more efficiently by using AI to identify exactly what a student should learn and when they should learn it. We provide adaptive and personalized curriculum to K-12, starting with English Language Arts, Math, and Writing.

As soon as a lawyer is made partner, they’re expected to bring in new business. However, it’s often not clear when a company is going to need legal counsel. So partners are encouraged to do business development with every person in their network in hopes that at least one of them generates new work. Osmaura gives partners a deeper understanding of prospective clients and legal support they might need at the right time. Our agents do this by monitoring the web, news, socials, and internal firm context to surface targeted matter opportunities for each partner.

Tolmo is an agent-driven security platform that secures every layer — code, CI, cloud applications at the speed of AI changes. Codebases are growing faster than ever, at a rate humans can’t keep up with. Tolmo deploys fleets of AI security agents on a live production knowledge graph that connects data from cloud services, containers, databases, and third-party providers (Observability, Security, etc.). Tolmo agents triage, verify impact, and remediate security issues autonomously, escalating to humans only when truly necessary.

AI agents are starting to act in the real world, but they can't reliably understand it. Geospatial data is complex, fragmented, and unstructured. Ask a model a specific question about a specific place and it guesses. Our API and MCP give agents cited answers about any location. Our mission is to index every inch of the earth, and make it as queryable as the web.

Linzumi is the chat your engineering team uses to direct dozens of AI coding agents. Kick off work, review it, ship it. All from chat. See every agent's work. Unblock the agents that need you from your laptop, your phone, the coffee line. Your teammates see it too. Every decision your team makes compiles into a source of truth - what to build, why it changed, how your agents should act. That same source generates new work and verifies completed work. Deliberative alignment, repurposed from aligning LLMs to aligning companies.

Glen is unified context for every agent and human on your team. Most MCP tools hand an agent a slice, one Slack thread, a single diff, and leave it to stitch the rest together itself. Glen reads from where your work already lives, code and PRs, issues, docs, and meetings, reconciles it, and gives your agents the full picture instead. With Glen a coding agent knows why a line was written, not just what it says. Ask a PR assistant anything about a change and the clarifying-question relay disappears. Bug bots read the intent behind the code before they comment, so they stop flagging deliberate choices as defects and stop repeating mistakes they already made. A sales agent can coach your head of sales off the full account history, how teammates handle similar calls, what was promised, where the deal stalled, etc. As the work gets done, Glen also distills it into skills and offers them back to the next agent, or person, who hits the same task.

Cerenovus is a company brain. We aggregate every kind of file a company produces — documents, PDFs, emails, Slack messages, spreadsheets, meeting notes — and convert them into a single markdown knowledge graph with native AI-agent integration. A company's information becomes uniformly readable both by humans and by the AI agents working inside the graph. On top of that foundation, Cerenovus maps the company as a system, drawing connections between the people, processes, and tools that make up real workflows. From that systems map, it infers where operations are inefficient and how to improve them. In practice, this tool will help executives make better decisions. Whenever a leader has to act on a question — how to restructure a team, whether to keep a vendor, or where handoffs between teams are breaking down — Cerenovus will give them an evidence-backed answer in minutes rather than weeks. Today, that work is done by consultants, by internal analysts, or by the executive asking around informally. Each of those approaches is slow, partial, and goes stale the moment it is delivered. Cerenovus produces the same answers faster, at a fraction of the cost, and keeps them live as the company evolves.

Zenbu is an extensible IDE for orchestrating coding agents. Shape it around exactly how you and your team work.

RealPact builds AI agents that handle the paperwork behind real estate transactions. They find property records like deeds, tax records, permits, parcel data, and MLS information, then use that data to fill out contracts, organize documents, track deadlines, and help brokerages close deals faster. We are starting by automating transactions and will expand into other parts of brokerage operations. Much of the software used across the real estate industry is outdated. With today’s advances in AI, we can build agents that handle this operational work and create modern software around how brokerages actually work. We currently sell to brokerages in New Hampshire and are expanding quickly.

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 280 other companies.

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 runs the entire drug program for biopharma teams, from discovery to FDA approval. Pharma runs on decades of closed legacy systems like Benchling, Veeva, Medidata, and SAS, with no AI layer and no MCP to connect agents to them. So we built that connection layer in one platform. Our agents run on top of their existing systems and data, with no migration, and do the work across the whole program: clinical evidence, trial design, statistical programming, and regulatory submissions. In just 3 weeks since launch, we're already being used by 500+ employees at top pharma companies, including Johnson & Johnson, AbbVie, Bristol Myers Squibb, Merck & Sanofi.

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.

Uno is the first third-party wallet granted access by Apple to become the default contactless wallet on iPhone. Uno is building a new payment rail, starting with a smarter wallet experience. The wedge is credit card points optimization. Founded by Saket Joshi (www.linkedin.com/in/saket-dattatray-joshi/) and Mordi Shadpour (www.linkedin.com/in/mordi-shadpour/)

We build modular humanoid robots and the software platform that allows anyone to train, deploy, and scale robots for real-world tasks. Today, teaching a robot a new skill requires robotics engineers, machine learning engineers, training infrastructure, and significant compute resources. We reduce that process to: Record → Upload → Train → Deploy. Our robots are modular and configurable, with interchangeable end effectors, configurable sensors, and multiple deployment configurations. We believe current humanoid embodiments are often overfit for specific use cases, so we're building a platform that adapts to different environments and applications. Long term, we want robotics to be as accessible as software development. Our goal is to become the default platform for training and deploying autonomous robots.

Light Anchor (YC P26) builds AI-run consumer brands that run at the speed of compute. Consumer brands have always been limited by the speed of humans: hiring, coordinating, and sleeping. AI agents that now run operations end-to-end break all those constraints, making a new kind of consumer business possible: one capped by compute, not headcount. Today, Light Anchor operates one flagship brand and multiple experiment stores, all run by AI agents on a shared platform that compounds with every new brand launched. The vision is thousands of consumer businesses, operating with little to no human intervention.

GodHands is a deterministic computer use layer that enables AI agents to reliably operate across browser and desktop apps. We make real-world workflows even on legacy or non-API systems more deterministic and predictable, without completely relying on fragile vision-based approaches. By leveraging our infrastructure, your agent can seamlessly operate across new or internal enterprise applications without requiring custom integrations or per-app handling.

Archer is the financial infrastructure for the AI data economy. We’re building global payouts, stablecoin banking, and yield infrastructure for the companies supplying data to frontier AI labs. As demand for high-quality data grows, millions of contributors around the world are generating, labeling, validating, and improving the datasets that power modern AI systems. Yet the financial infrastructure behind these platforms remains slow, fragmented, and expensive. Archer enables data marketplaces, AI labor platforms, and reinforcement learning environment providers to onboard workers globally, pay contributors instantly, and earn yield on idle balances. Today, companies use Archer to manage contributor earnings, global payouts, and treasury operations across hundreds of countries and payment corridors. We believe the next generation of AI companies will need purpose-built financial infrastructure. Just as cloud providers became the backbone of software, Archer is building the financial layer for the global workforce powering artificial intelligence.

Revnu is your AI growth team. Built to run your whole GTM while you focus on building, Revnu is the best way to grow with AI. Our agents run every growth channel: outbound, ads, SEO, social, etc all on one shared intelligence, so learnings from one channel automatically improve all the others. You keep shipping. We run the business.

Modern AI agents can send emails, change code repos, and talk with external APIs. Archal verifies your agent takes the right actions, and opens a fix PR when it doesn't.

The Imperfect app pulls your data to create a personalized AI coach for your next race. It adapts your training, recovery, and nutrition to how your body is actually responding.