July 22, 2026/1 min read
The mistake quietly killing most AI projects in private equity
Every AI use case in a PE firm has two scoreboards. Most firms only use one, and swapping them is what quietly kills good tools.
A framework that's been useful to me: every workflow in a PE firm has two distinct layers.
Mechanical work
- Assemble. Reconcile. Draft. Repeat.
- AI automates it. Judge it like infrastructure.
- Primary KPI: hours returned.
Judgment work
- Decide. Prioritize. Price. Challenge.
- AI augments it. Judge it like an analyst hire.
- Primary KPI: decision quality.
Both layers run through every desk, deal teams included.
The trap: swapping the scoreboards
- A judgment tool judged in hours saved looks like a failure, while it's changing which deals you see.
- An automation judged on the demo funds excitement over throughput.
The tools are fine. The scoreboards are swapped.
Before buying any AI tool, two questions
- Does it automate mechanics or augment judgment?
- What metric will prove it worked?
If the answer to either isn't obvious, don't buy it.
Originally shared on LinkedIn.
Originally shared on LinkedIn.