can ai make pay fairer?

Most companies know their gender pay gap exists. But very few measure it properly. That's increasingly not going to be an option — and Barbara Stolorz, co-founder of Warsaw-based startup Levelly AI, is building the tool that makes sure of it.

In the first episode of our She Shapes AI Award finalists series — ahead of the awards ceremony in London this April — Barbara joins Alice and Dan to talk about closing the gender pay gap, the EU legislation about to force the issue, and what happens when you show a board their own data and they tell you you're wrong.

The personal backstory

Barbara didn't set out to build a pay equity tool. She started with a different problem: compensation bands in job listings are almost meaninglessly wide, and nobody — especially women — ever really knows where they sit in them.

But a more personal frustration pushed her in a different direction. After returning from maternity leave twice, Barbara found her salary increases had simply been skipped. Not performance-related. Not explained. Just... not there — while inflation quietly eroded what she was earning.

It's a story that will be familiar to a lot of women. It's also, it turns out, highly measurable.

What Levelly AI actually does

The core product is less about AI than it is about speed and accessibility. The calculation underneath is, as Barbara puts it, "fairly simple mathematics" — comparing salaries across gender, nationality, disability status and other criteria, adjusted for role, seniority, and the full picture of compensation including bonuses and PTO. What would take an HR team hours (or days, depending on their Excel skills) takes Levelly seconds.

The AI layer sits on top of that: explaining what the charts mean in plain language, flagging the individuals and teams that need attention, and generating executive summaries that make it easier for HR to present findings to leadership. Next on the roadmap is an AI data assistant to help companies understand what data they actually need to provide — because right now, the data is fragmented, and that's often the biggest barrier.

The incoming legislation

The EU Pay Transparency Directive requires companies to publish their gender pay gap data by June 2027, with individual employees gaining the right to know where their pay sits relative to colleagues at the same level. European countries have until mid-2025 to implement it in local law.

Barbara estimates around 10–13% of companies currently have a clear picture of where they stand. The rest are only now starting to reckon with what "closing the gap" will actually cost — particularly for large employers, where a 20% gap across tens of thousands of employees represents a significant payroll adjustment.

"We don't have a gap."

One of the most striking moments in the episode is Barbara recounting Levelly's first commercial client. The team presented their findings — dashboards, data, a clear gap that couldn't easily be explained — to the board.

The board's response? We don't have a gap.

It's a moment that captures the whole challenge: the problem isn't always that companies can't measure pay equity. Sometimes it's that they don't want to know. The legislation is designed to change that calculation.

Beyond gender

Levelly's analysis isn't limited to gender. Barbara describes running calculations for nationality (including Ukrainian employees in Poland) and disability status — where she found the gap persisted even when employees with disabilities received additional PTO, because their base compensation still lagged behind. The tool can also model what gaps would look like in more mixed workforces, which matters in industries currently dominated by one gender.

Would a man have built this?

Barbara's answer is essentially: probably not, and the proof is in the room. When pitching at an early competition — with a panel that was mostly men — she was asked, with apparent genuine confusion, what a pay gap even was.

She's optimistic it's changing. But she's also clear-eyed: the AI industry is still male-dominated at the top, investment included. Her hope is that the accessibility of AI tools — many of them free — accelerates the pace at which women build, ship, and get recognised.

The thing that actually needs to change

The most quotable moment in the episode comes near the end, when Barbara makes a simple but sharp observation about language:

"When you think of a woman going to talk to her manager, she's going to kindly ask: may I have a salary increase? When I think of a man, it's: I'm going to ask because I deserve it. We do deserve it as well — and we should not kindly 'may I have.' It's: I deserve it. Please give it to me. And if not, what are the objective reasons why not?"

AI can close the measurement gap. But closing the confidence gap, and the cultural one — that's still on us.

Guest:Barbara Stolorz, Co-founder, Levelly.AI