YC opens Winter 2027 applications as its 2026 directory spans AI, software and biotech

Applications for YC’s Winter 2027 batch are open until November 2. This edition also reviews the breadth of YC’s 2026 company listings and Evidence’s hiring around agent-driven business intelligence.

Sunday, October 25, 2026

The latest from Y Combinator

YC’s Winter 2027 application window is open

A Y Combinator application notice says applications for the Winter 2027 batch are open until November 2. For founders considering YC, the deadline is the clearest time-sensitive development in the available source material.

Hacker News

Y Combinator

YC’s 2026 directory listings span AI, software and biotech

YC’s current directory pages list active 2026 companies across several areas:

SEO and AI discovery: Theo, OpenSEO and Armature are listed among seven SEO startups. Their products address small-business websites, AI-agent-driven SEO advice and visibility inside coding agents.
New York SaaS: The New York list includes 123 companies, including Agent 37’s persistent AI-agent sandboxes, Ekho Labs’ shipping-risk predictions, Marble’s AI restaurant operations platform and Bidflow’s AI-assisted electrical bidding.
Drug discovery: The directory lists 31 companies, including Atlas Discovery, Anto Biosciences, Paceline Bio and Ångström AI, spanning foundation models, clinical operations, molecular simulation and drug development.
San Francisco machine learning and consumer companies: The listings include newer companies such as Epsilab, Native, Saccade and Hundred, alongside established YC companies including Scale AI, DoorDash and Instacart.

The significance is breadth rather than a single announced program: the directory shows YC-backed activity ranging from AI-native business software and agent infrastructure to healthcare, biotech, logistics and consumer marketplaces. These pages are directory roundups, however, not announcements of launches, funding rounds or other specific company milestones.

Y Combinator — SEO startups

Y Combinator — New York SaaS startups

Y Combinator — drug discovery startups

Y Combinator — San Francisco machine-learning startups

Y Combinator — San Francisco consumer startups

Evidence is hiring around agent-driven business intelligence

A Software Engineer listing from YC-backed Evidence describes a product strategy built around Evidence Core, an open-source framework for reports and data products as code, and Evidence Studio, an enterprise BI platform combining an AI analytics agent with access controls, security and governance. The listing says the company’s open-source project has 7,000 GitHub stars and is used by thousands of teams.

This matters because it illustrates one YC company’s shift from traditional drag-and-drop BI toward code-native, agent-assisted analytics, while also showing that the company is expanding its engineering team and product surface.

Y Combinator — Evidence Software Engineer

Advice for early-stage founders

Treat the following as general operating guidance, not a report of Y Combinator news.

Start with the customer problem

Speak with potential users before committing to a solution. Ask about their current workflow, recent problems and what they already do to address them.
Look for repeated, specific pain rather than broad enthusiasm. Follow up on concrete incidents, workarounds and costs.
Keep notes on what you heard, separating observed behavior from requests for features.

Build the smallest useful product

Choose one user group and one important job to be done.
Build the narrowest version that helps someone complete that job, even if other parts remain manual.
Put it in users’ hands quickly. A working, limited product can reveal more than a long list of planned features.

Prioritize by learning and value

Rank work by the strength of the customer problem, the number of users affected and how quickly the work can produce useful learning.
Defer features that serve hypothetical needs or add complexity without improving the core outcome.
Review priorities regularly, but avoid changing direction in response to every isolated request.

Make distribution part of the product

Decide where the target users already seek information, solve problems or evaluate tools.
Test one or two repeatable ways to reach them, and measure whether those efforts produce qualified conversations, trials or continued use.
Turn early user interactions into a distribution loop: learn who benefits, refine the message and reach more people with the same need.

Use feedback to improve, not to outsource judgment

Watch how users use the product, ask where they get stuck and compare that with what they say they want.
Treat feedback as evidence about the underlying problem, not as an automatic specification.
Ship small changes, observe the result and keep a record of what changed. If adoption does not improve, revisit the problem, audience or value proposition rather than simply adding more features.