YC’s Fall 2025 Batch: Building the AI Economy’s Infrastructure
Share this article
The New Engine of AI Innovation
Y Combinator (YC) has long been a launchpad for the most ambitious tech ideas, but the Fall 2025 batch signals a shift in focus. With nearly two‑thirds of the 150 companies in the cohort dedicated to B2B software, YC is clearly betting on the infrastructure that will power the AI economy.
“These founders are building the rails for the AI era, not just riding it,” notes Forbes contributor Dasha Shunina.
The startups span a wide spectrum—from healthcare and fintech to industrial automation—but a common thread runs through them: the need to make AI safe, scalable, and seamlessly integrated into existing business workflows.
Zero‑Trust Meets AI
Multifactor tackles the most pressing risk in the age of autonomous agents. By offering zero‑trust authentication, authorization, and auditing for AI agents, it lets teams share online accounts with a simple link—securely.
The Gemini breach, dubbed the “invitation is all you need” exploit, highlighted how quickly agentic software can outpace traditional security measures.
With a focus on enterprise‑grade protection, Multifactor is positioning itself as the go‑to platform for organizations that need to trust AI without compromising compliance.
Giving Robots a Human Touch
Lightberry is turning humanoid robots from sci‑fi toys into practical teammates. By embedding vision, speech, and motion capabilities that mimic human interaction, Lightberry’s software can turn a remote‑controlled unit into an autonomous collaborator.
“We want robots to be coworkers and companions, not just tools,” founders Ali Attar and Stephan Koenigstorfer say.
The company’s work is timely as the robotics market pushes toward more autonomous, socially aware machines.
AI Agents That Remember
Hyperspell introduces a memory layer for AI agents, unifying context from Gmail, Slack, Notion, and Drive. Developers can give their agents persistent organizational memory, enabling more coherent multi‑turn interactions.
# Example: Attach Hyperspell memory to an agent
agent.attach_memory(Hyperspell.memory(['gmail', 'slack', 'notion']))
This capability is critical as enterprises look to embed AI into knowledge‑heavy environments.
From Unstructured Data to Structured Action
Unsiloed AI builds APIs that turn PDFs, slides, and images into structured data that LLMs can consume. With over 80 % of enterprise data in unstructured formats, the company is tackling a pain point that costs AI teams up to 40 % of their time.
“We’re building the infrastructure that makes documents as computable as databases,” says co‑founder Aman Mishra.
The Human Stories Behind the Tech
YC’s environment forces founders to iterate fast, talk to users nonstop, and focus on what truly matters. As one founder from Caddy explains, “People just want to get work done faster.”
“YC compresses months of growth into weeks,” says Unsiloed AI’s Mishra.
These narratives underscore why YC’s batch is more than a list of startups—it’s a cohort of people reshaping the way we build and use AI.
What This Means for the Industry
The batch’s emphasis on infrastructure—security, memory, data pipelines, and autonomous tooling—signals a broader industry trend: AI will be embedded, not added on. Companies that can provide the foundational layers will become the backbone of the next decade’s digital economy.
“If this batch is any indicator, the next generation of billion‑dollar companies won’t just use AI—they’ll define it,” concludes Shunina.
The Fall 2025 cohort is not just a snapshot of current trends; it’s a blueprint for the future of AI‑driven business. As these startups move from prototype to production, they will set the standards for security, scalability, and usability that every AI‑centric organization will rely on.
Source: Forbes, “The Top Startups To Watch From Y Combinator’s Fall 2025 Batch,” Dasha Shunina, 13 Nov 2025.