The Traceable Thesis

The near-term opportunity is real, narrow and unevenly evidenced. The near-term risk is concentrated in confidentiality, in claims made to investors, and in the quiet substitution of synthesis for evidence. A general partner deploying artificial intelligence is making three separate bets, and they fail in different ways. The first is that the technology will compress information work, which the available evidence supports. The second is that compressed information work will improve investment outcomes, which the available evidence does not yet show. The third is that the firm can say what it is doing without misdescribing it, and that is where enforcement has already landed. Artificial intelligence compresses the information work that surrounds an investment decision, and it does so most reliably in origination, diligence and portfolio monitoring. The compression is worth having. It becomes dangerous at the point where a general partner allows a synthesis to stand in place of the evidence behind it, because the fund then owns a conclusion nobody can reconstruct. The recommendation in this paper is to buy compression and pay for traceability, and to treat any workflow that cannot produce a source-linked record as unfinished, whatever its speed.