Harj Taggar on AI-native full-stack companies, selling to Fortune 500s, and YC's biggest batch trends

Summary

Harj Taggar, speaking from YC's offices during the W25 demo day, argues that the defining shift in the current batch isn't just faster growth — it's that early-stage companies are closing contracts with large enterprises and government customers faster than anything YC has seen before, driven almost entirely by AI.

Revenue patterns have changed shape. In SaaS, founders expected steady month-over-month growth. Now Taggar describes step-function trajectories: flat for a stretch, then a single large contract that leapfrogs months of organic growth. Some founders are walking into demo day having signed one significant deal that is enough to anchor a credible pitch.

Selling to enterprise — sooner than ever

The reason YC founders can land Fortune 500 or government contracts this early, Taggar argues, is structural. Incumbent engineering teams often don't believe in AI, which means large companies can't build these products internally. That creates a gap startups can fill directly, without first spending years accumulating smaller logos to prove reliability. The traditional playbook — sell to startups, build credibility, eventually approach enterprise — is no longer the only viable path.

Taggar still thinks the Stripe or AWS model, getting startups early and growing with them as they scale, is one of the most durable business models available. The point isn't that enterprise-first has replaced startup-first; it's that founders now have a genuine choice depending on what they're building.

AI-native full-stack companies

The more consequential batch trend is a move beyond vertical AI agents. A year ago the dominant pattern was infrastructure for building agents. The batch after that was vertical agents — AI applied to customer support, logistics, healthcare — sold to incumbents to improve operations. In the current batch, Taggar says a growing number of companies are skipping that sale entirely and becoming the operator.

Fernstone is building an AI-native insurance brokerage — not selling AI to brokers, but being the broker. Saba is doing the same in the trust and estate space. The bet is that AI lets these companies scale without the headcount that killed the full-stack startup model a decade ago. Taggar traces the original full-stack thesis back to a Chris Dixon blog post and the Spoon Rocket / Sprig era in San Francisco, where companies built mobile kitchens rather than just routing delivery orders. Those businesses ultimately couldn't scale without massive hiring, and marketplace models won. The AI version of the bet is that agents replace the labor, so the economics that broke full-stack startups no longer apply.