Zarna Pitches an AI Associate Class for Private Equity

Inside a private equity firm, the most numerate, sleep-deprived people in the building are usually the associates. They build leveraged buyout models, dissect confidential information memoranda, draft investment committee memos, update CRMs, and chase down the long tail of diligence questions that turn an idea into a deal. Zarna, a Y Combinator Fall 2025 startup based in San Francisco, is building AI tools for private equity deal teams designed to do that work – not by replacing the humans, but by giving every firm the equivalent of an always-on associate class trained on its own institutional knowledge.

Zarna describes itself as private equity’s first AI associate class. Its agents plug directly into a firm’s data – deal histories, prior models, IC memos, CRM records – and operate as if they were members of a deal team that never sleeps. In practice, that means Zarna’s agents read and analyze CIMs as they arrive, build first-pass LBO models that follow the firm’s own conventions, update CRM records with structured information drawn from documents and meetings, and draft IC memos grounded in decades of accumulated firm knowledge. The pitch to investment professionals is not that the agents will make better decisions than humans – it is that they will eliminate the rote work that today consumes the largest share of an associate’s week, freeing the deal team to spend more time on judgment, sourcing, and relationship-building.

The opportunity is unusually well-suited to AI. Private capital workflows are document-dense, repetitive, and governed by firm-specific conventions that compound in value over time – exactly the conditions under which a well-grounded AI system can outperform generic copilots. Firms have years of past models, memos, and decisions sitting in shared drives, and the institutional knowledge encoded in those artifacts is rarely searchable in any practical sense. By treating that corpus as the training and grounding context for its agents, Zarna positions itself as something closer to a firm-level reasoning layer than a productivity tool – an “agentic operating layer for private capital markets,” as one of the cofounders has described the long-term ambition.

Zarna was founded in 2025 by a four-person team: Rishabh Dhariwal, the company’s CEO; Vivan Agrawal; Hrishi Joshi; and Rakesh Mehta. The cofounders share a focus on the intersection of AI/ML and real-world capital markets workflows, and their product reflects a clear bet that the next wave of AI in finance will be vertical, deeply integrated with proprietary data, and tuned to the specific judgment cycles of investment professionals rather than generic enterprise users. Categorized under Investments and AI, Zarna joins a growing cohort of YC startups attempting to bring the cognitive load of capital markets professionals down by an order of magnitude.