AI Evolution Moves Family Offices From Query to Command

At the Family Wealth Report Fintech Forum in New York, Aleta CEO Ken Gamskjaer said AI has moved from answering questions to executing workflows — telling it to rebalance, flag tax implications and draft a memo, according to Family Wealth Report.

Gamskjaer said the infrastructure was not possible six months ago; since then OpenAI, Google, Microsoft, AWS and Anthropic have adopted MCP as a universal protocol for connecting AI to data, and Morningstar, PitchBook, LSEG and Aleta have launched MCP integrations, making private market intelligence, public markets data and portfolio analytics accessible to AI agents.

Truewind CEO Alex Lee said family offices' profile — small teams, complexity, recurring workflows and fragmented source data — is optimal for AI agents, but the first winning use case is supervised preparation, not autonomous decision-making. Humans must approve accounting treatment, resolve edge cases, and own final accountability.

Lee's assessment points to a practical near-term path: AI handles preparation work while humans make the final call. That is a step change from simple query answering, but stops short of handing decisions to agents.

For family offices with fragmented source data, the supervised model could reduce manual workloads in reconciliation and consolidation without surrendering accountability.

The MCP momentum is the notable infrastructure story here. Five major tech firms adopting the same protocol within six months, followed by MCP integrations from financial data providers, likely signals a push to make data accessible to AI agents without proprietary lock-in.

For family offices, the immediate implication is unglamorous: get systems into machine-readable shape before expecting agents to add value. Gamskjaer made that point explicitly, and Lee's supervised-preparation model reinforces it.

Expect family offices to prioritize data cleanup and machine-readable output before building custom AI agents, according to Gamskjaer's stated sequence.