Key takeaways from my conversation with Shubhi Rao, Founder and CEO of Uplevyl

This blog post provides a summarized version of my conversation with Shubhi Rao, featuring key highlights with slight alterations for readability. To listen to the full discussion, check out the latest episode of She Builds with AI on Spotify, Amazon Music, or your preferred podcast platform.

In this episode of She Builds with AI, I sat down with Shubhi Rao, Founder and CEO of Uplevyl, an AI company building secure women centered intelligence hubs that turn authoritative data into actionable insights, collaboration tools, and learning experiences.

Shubhi brings a rare combination into the conversation. She started her career as a software engineer, moved into finance and senior executive leadership, held roles at Ford, Tesco, PwC and Alphabet, and later served as CFO and COO of Dosh before its exit to Cardlytics. Today, with Uplevyl, she is working at the intersection of gender data, AI infrastructure, privacy, trust and practical decision support for women’s lives.

Our conversation centered on one of the most important questions in AI right now: what happens when the data behind intelligent systems does not fully reflect women’s lived realities?

AI is only as good as the data behind it

A key theme in our conversation was that AI is not neutral simply because it is technical. AI systems reflect the data, algorithms and assumptions behind them. If the data is incomplete, biased or built around a narrow default user, the results can become inaccurate, unfair or irrelevant.

Shubhi explained that this realization became especially clear to her while working on large, data heavy projects. The more she saw how much investment was going into AI, the more she asked a question that would eventually become central to Uplevyl’s work: where are the gender data sets?

For Shubhi, the issue is not that AI companies have deliberately excluded women. The problem is that many foundational systems are trained on data that reflects existing structures, and those structures have often made women’s experiences less visible.

This matters because AI does not only answer questions. Increasingly, it supports decisions, workflows, products, services and access to information. If the underlying data does not reflect women’s realities, the system may become faster and more powerful while still missing essential context.

What gender data really means

In the conversation, Shubhi emphasized that gender data is often misunderstood. It is not a single field that can simply be added to a model. It is the lived experience embedded in patterns of work, caregiving, money, health, safety, risk and major life transitions.

Women are not one homogeneous group. They show up as caregivers, mothers, employees, leaders, survivors, investors, patients and decision makers. Their experiences can differ by age, culture, race, health context, economic background, family structure and jurisdiction.

That complexity is exactly what women centered AI needs to hold.

A system that treats women as one generic segment risks flattening the very differences it needs to understand. True gender data has to account for different life stages, different constraints and different outcomes.

The life events mainstream AI often misses

Shubhi shared several examples of moments where women’s realities can differ significantly from the assumptions built into mainstream systems.

Motherhood can influence career breaks, financial planning, confidence, caregiving responsibilities and long term economic wellbeing. Menopause, breast cancer and heart symptoms that present differently in women show why health data needs more nuance. Divorce and widowhood can create legal, financial and administrative burdens that affect women in distinct ways.

One especially powerful example was widowhood and the great wealth transfer.

A widow may suddenly need to handle death certificates, estates, probate, Social Security, tax planning, accounts, insurance and advisors while also grieving. This is not simply a financial event. It is emotional, administrative, legal and deeply personal.

A generic tool may recognize the event. But a useful tool needs to understand the person living through it. What does she need in the first 90 days? What does she need in the first year? Which steps depend on her jurisdiction? What does she already understand about finance? Who else needs to be involved?

This is where AI can become incredibly helpful, but only if it is built through the lens of the person it is meant to serve.

Why Uplevyl works with organizations

Uplevyl is not only building a direct product for individual users. The company works with organizations that already serve women, such as employers, communities, financial institutions, advocacy groups, healthcare related organizations and other trusted partners.

Shubhi explained that many of these organizations already understand their communities and the problems they are trying to solve. What they need is secure, intelligent infrastructure that helps them serve those communities better.

This partner led model is important because it allows Uplevyl to stay in its lane as a technology company while working with organizations that bring domain expertise, community trust and real world context.

It also creates a different incentive structure. Instead of building a platform around data extraction, the focus is on safe infrastructure, curated knowledge, privacy and practical support.

Trust and privacy in sensitive AI applications

When AI touches rights, wealth, health, safety or major life decisions, trust becomes essential.

One principle stood out clearly in our conversation: AI should support a decision, but it should not make the decision for the person.

This distinction matters especially in sensitive use cases. Shubhi gave the example of supporting survivors of gender based violence as they try to understand workplace rights around paid leave, accommodations, unemployment insurance and discrimination. These rights can be fragmented, complex and written in legal language that is difficult for everyday people to understand.

AI can help make this information more accessible. It can help someone understand what rights may apply, what forms may be needed and what steps could come next. But it should also protect the user’s privacy, create psychological safety and point people toward human experts where needed.

For Shubhi, privacy first AI means enabling anonymity, minimizing data retention, deprecating chat history where appropriate and not using sensitive interactions to retrain models for commercial purposes.

In high stakes situations, respect for women’s data is not a feature. It is the foundation.

Building women centered AI without oversimplifying women

Another important theme was the challenge of building women centered AI without treating women as one group.

Shubhi shared that Uplevyl had to build first party data sets because gender data sets cannot simply be purchased off the shelf. The company brought in domain experts across areas such as medicine, law, psychology, board leadership, wealth and other fields to help create a richer knowledge base.

This matters because women centered AI must reflect many types of expertise and many types of lived experience. It has to consider not only gender, but also culture, race, age, geography, economic context and the specific problem being solved.

The goal is not to create a narrow version of AI for women. The goal is to build intelligence infrastructure that can understand complexity instead of erasing it.

From critique to building

One of the strongest messages from the episode was Shubhi’s call for women to actively shape AI.

She was clear that raising concerns about bias is important. The issues are real and need to be named. But critique alone is not enough, especially when AI is moving so quickly.

Women need to be part of the solution. That does not mean every woman needs to become an engineer or build a foundation model. It means women need to use their voice inside companies, communities, boards and product teams.

They can ask where the data comes from. They can ask who the system was tested on. They can ask what assumptions are built into the workflow. They can ask how privacy is protected, how risk is governed and who might be harmed if the system gets something wrong.

Building AI is not only writing code. It is also defining the problem, shaping the product, questioning the data, protecting the user and deciding what kind of future the system is allowed to create.

Where women have leverage now

Shubhi pointed out that almost every function inside organizations is being affected by AI driven process redesign. This creates risk, but also opportunity.

Women in HR, product, marketing, finance, operations, legal, strategy, leadership and board roles can all influence how AI is adopted. The questions may look different depending on the function, but the underlying responsibility is similar.

Where does the data come from?
Who benefits from the system?
Who is included in testing?
What happens when the model is wrong?
What risks are legal, reputational or operational?
What kind of governance is needed?
How do we make sure the tool serves people instead of simply optimizing a process?

As AI becomes embedded in workflows and decision making, these questions are no longer abstract. They are part of responsible leadership.

Making AI accessible

Toward the end of the conversation, Shubhi spoke about accessibility and upskilling.

AI tools are becoming more powerful and more complex at a rapid pace. What once felt advanced quickly becomes the baseline. For women, this creates an urgent need to build fluency and confidence.

The goal is not for every woman to become deeply technical. The goal is for more women to understand how to use AI, how to question it, how to guide it and how to participate in the rooms where decisions are made.

Otherwise, the gap between those who can shape AI and those who are shaped by it will continue to grow.

Final thoughts

What stayed with me most from this conversation is that AI is not only about technology. It is about the data of our lives.

It is about what gets seen, what gets ignored and what becomes embedded into products, services and decisions. If women’s lived realities are missing from the data, AI can become more efficient while still failing to understand half the world properly.

But Shubhi’s message is also deeply hopeful.

We can still shape this.

We can build better products, ask better questions, design for trust, protect sensitive data and bring women’s experiences into the systems being created right now.

The future of AI will not become inclusive by accident. It will be shaped by the people who decide to build it that way.

Listen to the Full Episode

This conversation with Shubhi Rao is full of thoughtful insights on gender data, women centered AI, trust, privacy, product leadership, responsible AI, and the future of intelligent systems that truly understand the people they are meant to serve.

If you are a woman building with AI, a leader responsible for AI adoption, a founder designing products, or someone thinking about how technology can better reflect women’s lived realities, this episode offers a grounded and deeply relevant perspective.

Listen to the full episode of She Builds with AI and share it with someone who should be part of the AI conversation now.

Because building better AI is not only about more powerful models.

It is about making sure women’s lives, data, questions and leadership help shape what comes next.

 ✨ Stay Connected with Shubhi Rao / Uplevyl:

🔗 LinkedIn Shubhi Rao: https://www.linkedin.com/in/shubhirao/
🔗 LinkedIn Uplevyl: https://www.linkedin.com/company/uplevyl/
🔗 Instagram Uplevyl: https://www.instagram.com/uplevyl/