Metal’s Co-founder and CEO Taylor Lowe and Court Square’s Resource Partner Jessica Pizzo on building the data layer for AI, in PE Hub

Before a private equity firm decides whether to buy or build AI infrastructure, it needs a coherent data foundation. As Lowe puts it: "Investing is about reaching conviction. Arriving at a decision requires data."

Pizzo shared Court Square’s approach to AI as a long-term strategic initiative, emphasizing data quality and the integration of AI into existing workflows. By using AI to evaluate opportunities against the firm’s investment criteria, Court Square aims to free its investment team to focus on the most promising opportunities.

Key Takeaways

• AI starts with the data. Whatever path a firm takes, without a coherent data foundation it will run into walls.

• The AI stack has three layers. A data layer holding the firm’s deals, portfolio and decisions; an intelligence layer for search, agents and deal scoring; and an interface used by both humans and AI agents.

• A proof of concept is not a scaled system. Serving 50 to 100-plus professionals with firm-wide data is a different challenge.

• Run AI as a program, not a project. It needs a standing owner with a rare mix of AI, data and deal-team fluency.

• Avoid model lock-in. The context layer should be built to absorb frontier model upgrades, so firms can switch models as the leader changes.

Read the full PE Hub interview: Metal on how PE managers can lay the data foundations →