YC Request for Startups 2025 — Complete Breakdown

This page breaks down every 2025 RFS category across all four editions, what each one actually means for founders, and how to use the RFS to position your application.

YC is explicit about this on the RFS page itself: these ideas represent a fraction of what YC funds. You do not need to work on an RFS category to apply or get accepted. The RFS is validation, not a prerequisite. What it does tell you is where partners are especially receptive — meaning an application in an RFS category enters the review process with a partner already primed to be interested.

The RFS is also a leading indicator of where YC has spotted an inflection point. Categories appear on the RFS when partners believe the technology, market, or regulatory environment has shifted enough that a startup starting today has a real shot at something that was not possible 12-18 months ago.

Winter 2025 RFS — Key Categories

The Winter 2025 RFS focused heavily on AI infrastructure and the next layer of AI application development. Key categories included:

Full-Stack AI Companies

YC's call for startups that use AI not as a feature but as the entire operational spine — replacing human workflows wholesale rather than augmenting them. The thesis: AI is now capable enough that a 2-person team can operate what previously required a 50-person company. YC is looking for founders who pick a legacy industry and rebuild it from the ground up using AI-native operations.

Voice AI

YC explicitly called for startups building Voice AI products across customer service, healthcare, education, and consumer applications. The inflection point: voice AI quality crossed a threshold in 2024-2025 where synthetic voices became indistinguishable from human voices in most contexts, opening categories that required human voice as a prerequisite.

AI for Scientific Advancement

Startups applying AI to accelerate drug discovery, materials science, climate modeling, and other scientific domains where the bottleneck is data analysis and hypothesis generation rather than experimental throughput. YC is specifically interested in companies that shorten the loop between hypothesis and validated result.

AI Personal Assistant

Not another chatbot — YC wants an AI that knows you specifically: your schedule, preferences, relationships, ongoing work, and goals, and proactively acts on your behalf rather than waiting to be prompted. The differentiation from existing products: longitudinal context and genuine proactivity.

Healthcare AI

Specific to the Winter 2025 edition, YC called for AI that reduces the administrative burden on clinicians (documentation, prior authorization, coding) and AI that improves diagnostic accuracy in specific high-volume, high-error-rate conditions.

AI for Education / AI Personal Tutor

Adaptive, individualized learning that provides the quality of 1:1 private tutoring at consumer scale. YC is looking for products that measurably improve learning outcomes, not just engagement.

Software Tools to Make Robots

The tooling layer for robotics development: simulation environments, training infrastructure, deployment frameworks, and testing tools that reduce the time and cost of building and deploying physical robots.

AI Residential Security

Consumer and SMB security products that use computer vision and AI to replace the monitoring model that depends on human operators watching camera feeds.

Spring 2025 RFS — Key Categories

Spring 2025 RFS continued themes from Winter 2025 while adding several new areas reflecting the rapid pace of AI development.

Secure AI App Store / AI OS Layer

YC called for a new kind of AI application distribution layer — one that handles user data privacy, shared memory across AI apps, developer infrastructure, and curated quality control. The thesis: the current model of isolated AI applications that cannot share context or memory is a structural limitation that a new platform layer can solve.

Datacenters

YC explicitly called for startups building data center infrastructure — power, cooling, materials procurement, project management, and construction software. The thesis: demand for compute is growing faster than the traditional data center development pipeline can support, creating a structural supply shortage that represents a large commercial opportunity.

Next-Generation Construction

AI-powered construction technology: planning, project management, materials optimization, and on-site coordination. The specific YC framing: construction is one of the least digitized major industries by revenue, and AI makes a step-change in productivity newly possible.

Government Software

AI-powered automation for government administrative tasks. The specific opportunity: government agencies perform massive volumes of routine administrative work that is well-suited to AI automation but has not been addressed by enterprise software because of the complex procurement and compliance environment.

Summer 2025 RFS — Key Categories

Summer 2025 built on previous editions with emphasis on full-stack AI companies and the emerging agentic infrastructure layer.

Full-Stack AI Companies (continued emphasis)

Startups that build AI-native operations to compete directly with established service businesses — accounting, legal services, insurance brokerage, healthcare administration — by replacing the human service layer with AI rather than building software tools for humans to use.

Multi-Agent System Infrastructure

The infrastructure layer for deploying fleets of AI agents — orchestration, memory, state management, error handling, security, and monitoring. As individual AI agents proved capable, the bottleneck shifted to managing multiple agents working together reliably.

AI-Native Enterprise Software

The "Cursor for [vertical]" thesis: building AI-native replacements for established enterprise software categories where the incumbent products were built for human workflows and cannot be meaningfully improved by adding AI as a feature.

Video Generation

AI tools for creating, editing, and producing video content. The inflection point: video generation quality crossed a threshold in 2024-2025 that made it commercially viable for a range of production use cases beyond novelty applications.

Fall 2025 RFS — Key Categories

Fall 2025 emphasized workforce transformation, AI in the physical world, and the 10-person company thesis.

Retraining Workers for the AI Economy

Platforms that help workers displaced by AI automation retrain for roles that remain human-led or that emerge from the AI transition. YC framed this as one of the largest social and commercial challenges of the next decade.

The 10-Person $100 Billion Company

YC's most ambitious thesis: AI makes it possible for extremely small teams to build and operate companies at scales previously requiring hundreds or thousands of employees. YC is looking for founders who take this seriously as a design constraint.

Physical World AI

AI that operates in and improves the physical world: manufacturing, field service, agriculture, construction, and infrastructure. The transition from AI that lives in software to AI that acts in physical environments.

Across all four 2025 RFS editions, six themes appear consistently — indicating the categories YC is most committed to:

Themes that appear in only one 2025 edition:

• AI Residential Security (Winter only)
• Retraining workers (Fall only)
• The 10-person $100B company (Fall only)

Single-edition themes are typically newer ideas YC is exploring — not necessarily less important, but less tested as investment theses.

Use it to validate, not to fabricate

The worst application mistake inspired by the RFS is building a company around an RFS category because YC listed it — rather than because you have genuine insight and founder-problem fit in that space. Partners can immediately distinguish a founder who pivoted to match the RFS from a founder who was already building in the space when the RFS validated their thesis.

Cite it explicitly if your company fits

If your company genuinely fits an RFS category, say so in the "Why YC?" field: "Our product directly addresses YC's Winter 2025 RFS call for Healthcare AI that reduces administrative burden — specifically the prior authorization problem we have been working on for 18 months." This is not pandering; it is relevant context that helps a partner immediately locate your application within their investment thesis.

Use the language of the RFS

The specific framing YC uses in the RFS reveals how partners think about these problems. Reading the actual RFS text and using its conceptual framing in your application signals that you understand what YC is actually looking for — not just what the category sounds like from the outside.

Multiple editions signal conviction

If your category appears in three or four of the 2025 RFS editions, that is a signal of strong, sustained YC conviction. If it appears once, it is a signal of emerging interest. Both are useful — but the multi-edition themes carry more weight.