AI investment strategies are flooding the family office world…with mixed results

For better, or worse, family offices need to cope with a flood of AI strategies seeking to win influence and fees over the way they invest. 

The latest deal involves Inference Research of Hong Kong, which wants to use AI to build trading systems. It has raised $20 million in seed capital from Avenir Group, which is backed by a family office owned by crypto entrepreneur Li Lin.

A former Oracle engineer, Li Lin (aka Leon Li), developed and sold Huobi (now HTX), a large Seychelles-based crypto exchange. Last year, Avenir owned $690 million of bitcoin ETFs, making it one of Asia’s biggest holders. Inference Research wants to use its capital to achieve asset pricing and liquidity provision across markets. 

According to Avenir: “This approach aligns closely with our mission to bridge traditional finance and digital assets.”

AI investing strategies have recently performed well. Minotaur Capital of Australia has beaten the MSCI index by 6.4% a year since its inception in 2024. A core strategy from Smart Wealth of Switzerland has outperformed its peers, with a multi-asset strategy return of over 41% over five years, though it has lagged the MSCI equity index.

Jim Simons’ Renaissance Technologies, launched in 1988, became the best tech-driven manager in history by producing an average annual return of 66% over three decades, using algorithms to spot a massive number of mispriced situations. 

Algorithms now fuel a large proportion of stock-market trading, led by high-frequency trading, in which deals can be processed in less than 100 nanoseconds to beat the competition. 

Several family offices believe in Altruist, led by Jason Wenk, once the youngest professional employed by Morgan Stanley

A newly published Harvard Business School report has found that more sophisticated AI can replicate 71% of decisions made by active managers. AI cannot mimic all the factors which produce outperformance or underperformance. But it can interpret market flows, trends, and peer-group analysis, suggesting it can play a big role in portfolio analysis and strategy.

Passive strategies, enabled by technology, have already undermined the position of active managers, and more advanced AI providers are now out to steal further market share from active managers and wealth advisers, whose fees for portfolio administration and stock tipping are lucrative.

In February, shares in wealth advisers were rocked when Altruist Corporation announced an AI tool that can offer personalised advice by scanning tax stubs, account statements, and other documents. This could undermine fees from big books of business retained by the adviser. Among other services, Altruist offers speed, promising to connect family investors with advisers in under 60 seconds. 

Several family offices believe in Altruist, led by Jason Wenk, once the youngest professional employed by Morgan Stanley.

It has raised $600 million from the likes of Venrock, Iconiq Growth, Vanguard, angel investor Marty Bicknell of Mariner Wealth and a family office owned by iconic financial adviser Ron Carson, now embracing an alternative lifestyle, plus the name Omani, after embracing psychedelics.  

Another AI adviser, Avantos, co-led by former McKinsey partner Bassam Chaptini, has raised $35 million from established financial companies, including Bessemer Venture, Guardian Life, Vanguard, SEI, MIT’s E14 fund, and consulting firm Mercer.  It wants to provide client service, an area where expertise is expensive to hire. Guardian is now a client.

In early March, Addepar, an adviser to family offices led by Joe Lonsdale, a Peter Thiel associate, launched Addison, a process which uses AI to deliver intelligence across client portfolios: “To streamline data operations, client intelligence and client management workflows, from data remediation to analytics, forecasting and reporting.” Addison won’t be making investment decisions, but it certainly aims to inform them.

AI giant Anthropic, on track to deliver overall revenues of $19 billion this year, up from $9 billion in 2025, has unveiled a package for AI wealth solutions. 

Its Claude CoWork plug-in agents will be able to perform automated portfolio and tax analysis, suggest portfolio rebalancing and execute the trades. Norway’s $2 trillion sovereign wealth fund, run by Norges Bank, uses Anthropic to screen its portfolio for reputational and ethical risks.

BlackRock routinely capitalises on institutional hunger for quant systems. It has developed a stream of AI offerings, some of which have emerged from its pioneering Aladdin trading systems.

DeepVest is an interesting company, led by Toby Wade, which sees a big future in providing advice to investors, including family offices trying to find a way through the AI thicket. It is marketing a product called AdvisorLab, which uses underlying data to help check portfolios for hidden fees, tax inefficiencies, portfolio concentration, asset-allocation misalignment, and rebalancing. 

Large investors are using DeepVest, and others are developing bespoke solutions. According to Dell Technologies, a leading European trading house is investing $1.2 billion in data centres that support AI-driven forecasts for 50,000 financial instruments and in building solutions that can be scaled to handle data growth and speed. 

DeepVest is sceptical about the ability of AI large language models, which evolved from text predictors, to become massive research engines. In February, it published a white paper that said big AI providers can use the wrong data sources, hallucinate numbers, fail to compute answers, and offer results they cannot repeat.

The white paper concludes: “General-purpose AI tools are not suitable for professional investment work where accuracy, repeatability and regulatory compliance are required.”

The market’s febrile state could also challenge AI. President Donald Trump’s war against Iran, which could end up sucking capital out of the market, driven it to all-time highs by US exceptionalism. And AI could be poorly positioned to cope because, out of necessity, it relies on backwards-looking data.

A capital shortage triggered by the Iran war may (or may not) undermine big momentum trades and passive managers with a large weighting in expensive tech stocks. Private assets are eroding market transparency, exacerbating fears of a large debt bubble beneath the surface, as the value of private equity funds continues to crater. 

All may, of course, be well. Efficiencies delivered by AI, plus central bank fiscal easing, could give values a boost if the Iran war ends soon. 

But it’s hard to see how AI can make sense of a sudden market implosion. We could yet see a rerun of the ‘quant quake’ of 2007, when investment models froze, and hedge funds crashed amid the weird market gyrations that preceded the great financial crisis.