AI Fundamentals for LPs: Structured vs Unstructured Data

What LPs & Allocators need to understand about structured vs unstructured data

There are two types of data every LP team works with and understanding the difference is the foundation for understanding where AI can genuinely help.

Structured data lives in your systems: returns, allocations, exposure tables, etc. It’s organized and searchable. Technology has historically handled it decently.

Unstructured data is everything else. For LPs & allocators, that means manager meeting notes, reference calls, internal discussions on investment themes or asset allocation, etc. This is where most of your real institutional knowledge sits, but it has historically been inaccessible to technology. It lives in documents, email threads, and people’s heads.

The consequence? When a team member leaves, a lot of that knowledge leaves with them. There’s no system adequately capturing it, no way to query it, no institutional memory.

Large language models are specifically built to read and surface meaning from this kind of data. Which means the knowledge your team has built up over years of manager meetings, calls, and decisions can finally become a living, queryable asset.

Before implementing any AI system internally, you should prioritize how you will facilitate AI system access to both your structured and unstructured data. Without that foundation, even the most sophisticated AI tools will only ever work with a fraction of your team's actual knowledge - leaving the most valuable institutional intelligence untouched.