“Despite AI's potential to enhance sectors like healthcare, education, and business, it often mirrors reality and its societal prejudices and can manifest itself through unequal treatment in hiring decisions, academic recommendations, or healthcare diagnostics, systematically disadvantaging women.”

What all this means is that while AI is a marvel and use cases multiply in sectors such as private banking and wealth management, users must think carefully about how they employ AI. Critical reasoning, alertness and curiosity among flesh-and-blood humans remain vital. 

These issues are front of mind for Shubhi Rao, founder and CEO of Uplevyl, an AI company in the US that builds vertical knowledge layers through a gender lens, relevant for workplace rights, financial decisions and reproductive health. Uplevyl was founded in 2021. Rao, based in Austin, Texas, argues that there is a hidden gender gap in AI adoption that influences financial advice, wealth building, and access to quality planning.

Women's financial plans often fall short, and the wealth management sector must adapt, she told Family Wealth Report in a call. One factor is that many financial plans tend not to capture that women typically outlive men, requiring investment to accommodate this.

“AI is as good as the datasets that power it. If you think of frontier models…the datasets powering these models are scraping from the internet. None of them are seeing what we should use for specialized applications,” she said. “You tend to be heavier on the advisor's side of the table on their services than on the investment side.”

Rao brings experience from a number of fields, including economics. She is also an honorary advisory council member of the Federal Reserve Bank of San Francisco, and has served as an Economic Council board member of that institution. She is also a contributor to Forbes magazine. 

Assumptions in the data
A Cerulli survey of affluent investors released in in September 2024, for example, found that 50 per cent of affluent women investors rely heavily on an advisor, against 34 per cent of men. Women face higher lifetime healthcare costs in retirement, largely because they live longer, she said, and this will reflect in how they think of finances. HealthView Services' 2026 report puts women's retirement healthcare costs at about 15 per cent higher than men's on average. 

“At the moment, AI models don't reflect this,” Rao said. 

Biases by humans can be multiplied by AI. 

In 2019, an article by Kings College London referred to research from Dr Ylva Baeckstrom of King’s Business School. It found that “male and female financial advisors make judgements about wealthy clients based on their own and their client’s gender and give different recommendations as a result.” Dr Baeckstrom, a former private banker, based her research on her work with 129 UK-based investment advisors specialising in HNW clients.

Switching back in the US, a report this year from Stanford and MIT found that AI chatbots give women 3 percentage points less equity exposure, costing them roughly $60,000 in simulated retirement wealth by age 60. The team behind AI Financial Advice: Supply, Demand, and Life Cycle Implications, released as MIT Sloan Working Paper 7377-26 in May 2026, simulated the life of a virtual saver following advice from GPT-5.2 and Gemini 3 Flash. 

Such results show that using AI models in wealth management must be done carefully, Rao said. 

“The MIT and Stanford study found that after a job loss the models recommend cutting spending by about as much as income falls, and that more than 98 per cent of recommended retirement withdrawals sit at or below a 4 per cent rule,” she said. “None of these studies test widowhood or other life-event sequences. A widow's first year is a series of dated, partly irreversible decisions that depend on where she lives, and that is where we focus.”

“These are systemic issues,” Rao said. “There is limited data on gender differences in finance to power algorithms at the moment, and we have not found a published benchmark for life-event guidance.”

Her comments come as the wealth industry, still a relatively male-dominated one, needs to adjust as more women inherit and create wealth. This news service has covered such issues from a variety of perspectives, as seen here and here.

What is to be done?
Rao argues that AI it vacuums data from the open internet, which is a sort of horizontal model. Instead, there is a need for the technology to be placed in more vertical “closed” structures or “multi-tenant” ones. 

“Closed means the guidance is drawn from a curated, source-verified body of rules that we maintain, with no reliance on what a model absorbed from the open internet,” Rao said. “Multi-tenant means each firm that uses the platform is its own tenant, with its own configuration and boundaries. The system is stateless, so it holds no client data between calls.”

Life events
As part of this approach, Uplevyl is mapping the rules behind the decisions that each life event triggers for women, starting with widowhood across US jurisdictions, for example. 

“Each life event sets off a different chain of decisions, deadlines and rules, and which rules apply depends on where a person lives and what accounts she holds. We are mapping the rules that decide what a person must do, and by when, after a life event, for the jurisdiction they live in,” Rao said. “The list of life events is open-ended. Widowhood comes first because it is the hardest. A surviving spouse can face survivor benefit claims, inherited retirement account elections, a change in tax filing status, Medicare premium adjustments, life insurance claims and state probate rules within months, each with its own deadline, and some choices cannot be reversed.”

“The platform is designed to take the life event and the jurisdiction and return the decisions in sequence, with deadlines and flags for when a tax, legal or insurance specialist is needed. The advisor stays in charge of the relationship. The architecture extends to divorce, retirement, disability and other life events.

“We need to create workflows that are specific for advisors…frontier AI models are not designed for it,” Rao continued. 

“In our conversations with advisors, women tend to use more of the advisor's time on life events,” she said. 

Uplevyl has a team of 25 and four advisors, and its clients are a combination of enterprise and non-profit organizations. Its knowledge layers cover US jurisdictions.

Profitability under pressure
Beyond the gender issue, Rao talked about the pressure that wealth management is under, in terms of profits. 

“Globally, profitability per AuM is declining,” she said. There is an advice shortage, and the younger generation is less interested in following a financial advisory career path. A lot of those who enter the field tend to leave, she said. “It looks as if a perfect storm is coming.”

As evidence, Rao cited the 2025 Global Asset & Wealth Management Report of PwC which surveyed 300 firms in 19 countries. If found that profit per dollar of assets under management fell 19 per cent since 2018, with a further 9 per cent decline expected by 2030. Expenses absorb 68 per cent of each dollar. 

The picture for advice shortages varies by market. In the US, McKinsey estimates a shortfall of roughly 100,000 advisors by 2034, within a range of 90,000 to 110,000. In the UK, the Financial Conduct Authority has said that advisor numbers have held at around 31,000 since 2021; the average advisor is in their late 40s. 

The challenge of profitability was highlighted recently in the UK by BWC Benchmarking in its annual study of performance of the UK sector, pointing to relatively low net organic growth once M&A activity is taken out. 

The stakes
Rao said the importance of ensuring AI works for all sides of society is more important than ever. She noted how in May, for example, Schwab's chief executive, Rick Wurster, said he sees AI becoming a front door for clients, including those below the $1 million needed for a dedicated advisor. This means that generative AI is about to sit between firms and a much larger group of clients. 

A Harvard Business School working paper drawing on more than 300,000 respondents estimates generative AI adoption at 47.8 per cent for men and 39.3 per cent for women.