In a bold and defining move, Y Combinator (YC) unveiled its Spring 2025 batch on June 15, and what stands out is more than just the number of startups. A remarkable 70 startups—nearly half the cohort—are building agentic AI systems, and each of them has been granted $500,000 in funding. That’s not just an accelerator announcement; it’s a clear signal: agentic AI is no longer a buzzword—it’s the future YC is betting on.
The term “agentic AI” refers to artificial intelligence systems designed to act autonomously—making decisions, taking initiative, and executing tasks without needing continuous human input. These AI agents are not just tools; they function more like digital employees or assistants, capable of completing workflows, managing tasks, or even negotiating on behalf of users. And this spring batch shows that Y Combinator wants to lead the charge.
This is YC’s first-ever Spring batch, and the fact that nearly 50% of the selected startups are focused on agentic AI underlines a strong strategic shift. These startups aren’t just building clever tools; they’re creating autonomous agents that can navigate complex systems—whether it’s filing insurance claims, securing mortgage approvals, or even training robots through voice prompts.
Among the standout startups from the Demo Day presentations was Aegis, which is working on automating insurance appeals in the healthcare space. By interpreting dense insurer policies and crafting responses to denied claims, Aegis aims to act as an AI-powered advocate for doctors and clinics, reducing administrative burden and accelerating payouts.
Then there’s Approval AI, a fintech solution that helps users through the mortgage process. Rather than just offering rate comparisons, Approval AI takes a proactive role—negotiating terms, preparing documentation, and ensuring that users land the best possible deals without wading through financial jargon.
Another noteworthy name is Mbodi AI, which focuses on robotics. Traditionally, teaching robots new tasks involves manual programming and loads of time. Mbodi flips that script by letting users train robots using natural language. Tell it to “pick up the blue cup,” and the robot learns the skill through that command—a game-changer for industries looking to scale automation quickly.
Health and wellness also saw a presence with Galen AI, which pulls data from wearables, health apps, and medical records to create a personalized wellness advisor. It doesn’t just track steps or calories; it analyzes trends, gives recommendations, and adjusts goals—all autonomously.
YC’s selection also featured agentic infrastructure tools like Docket, which helps developers test AI models autonomously across web applications, and Casco, focused on vulnerability scanning for AI systems. These tools are crucial for developers building the very agents YC is banking on, forming the underlying scaffolding for the agentic future.
This year’s batch also marks a generational leadership shift at YC. Longtime partners Dalton Caldwell and Michael Seibel have stepped into emeritus roles, passing the torch to Jon Xu and Andrew Miklas, who were recently promoted to general partners. Their background and fresh perspectives seem to align with YC’s AI-first vision. Under their guidance, the Spring 2025 batch clearly signals where startup innovation is headed next.
Beyond individual startup innovations, this agentic-first batch reflects larger industry trends. Investors have been heavily backing AI agent startups in recent months. For example, Qevlar raised $14 million for agentic security automation, Artisan secured $25 million for business process agents, and HoneyHive raised $7.4 million to build tools that test and optimize AI agents throughout their lifecycle. Y Combinator’s batch appears strategically placed within this growing market momentum.
Each of the 70 agentic startups now begins a three-month sprint in San Francisco, receiving YC mentorship, community support, and a stage to refine both their technology and business strategy. As they prepare for the final Demo Day at the end of summer, all eyes will be on whether these AI agents can go beyond prototypes to secure customers, attract follow-on funding, and deliver tangible results in the real world.
From healthcare and finance to infrastructure and robotics, YC’s Spring 2025 cohort isn’t just dabbling in AI—it’s doubling down on autonomous, decision-making systems that may very well define the next phase of software innovation.
With its clear focus, strong backing, and a growing market eager for such tools, Y Combinator’s latest batch is more than an accelerator program—it’s a launchpad for the age of AI agents.
Key Facts at a Glance
| What | Details |
|---|---|
| When announced | June 15, 2025 |
| Number of startups | 70 (out of 144 total) |
| Focus area | Agentic AI (autonomous decision-making systems) |
| Funding per startup | ~$500,000 |
| Program duration | April–June 2025 (3 months in SF) |
| Notables at Demo Day | Aegis, Approval AI, Mbodi AI, Plexe |