Does It Tie Out? Towards Autonomous Legal Agents in Venture Capital
Pierre Colombo, Malik Boudiaf, Allyn Sweet, Michael Desa, Hongxi Wang, Kevin Candra, Syméon del Marmol
2025-12-23
Summary
This paper focuses on the difficult task of 'capitalization tie-out' which is a crucial step lawyers take when companies are trying to get funding from venture capitalists. It's about making sure all the details of who owns what in the company are perfectly accurate and backed up by legal documents.
What's the problem?
Currently, even advanced AI systems struggle with capitalization tie-out. This is because it requires carefully analyzing many different legal documents at once, being able to pinpoint exactly where information comes from, and always giving the correct, definitive answer. Existing AI often makes mistakes or can't clearly show *why* it arrived at a certain conclusion, which isn't acceptable when dealing with legal ownership.
What's the solution?
The researchers identified capitalization tie-out as a good test for legal AI and tested existing AI systems. They then proposed a new AI architecture, called a 'world model,' specifically designed to handle this kind of complex legal work. This new system aims to be more reliable and provide clear evidence for its conclusions.
Why it matters?
This work is important because automating capitalization tie-out could save lawyers a lot of time and reduce errors. More broadly, it's a step towards building AI that can truly assist with complex legal tasks, going beyond just answering simple questions and actually helping with real-world legal workflows.
Abstract
Before closing venture capital financing rounds, lawyers conduct diligence that includes tying out the capitalization table: verifying that every security (for example, shares, options, warrants) and issuance term (for example, vesting schedules, acceleration triggers, transfer restrictions) is supported by large sets of underlying legal documentation. While LLMs continue to improve on legal benchmarks, specialized legal workflows, such as capitalization tie-out, remain out of reach even for strong agentic systems. The task requires multi-document reasoning, strict evidence traceability, and deterministic outputs that current approaches fail to reliably deliver. We characterize capitalization tie-out as an instance of a real-world benchmark for legal AI, analyze and compare the performance of existing agentic systems, and propose a world model architecture toward tie-out automation-and more broadly as a foundation for applied legal intelligence.