In 2026, AI is no longer a nice-to-have for founders — it's an important part of how teams are structured, products are built and businesses plan their growth. But as the technology matures, a new question is emerging. What does it mean to launch and scale a startup globally in the age of AI? And what does a genuinely ‘hybrid’ workforce of humans and agents look like in practice?
At our first-ever rooftop event in London, Wise Business’ Hannah McGrath led a conversation with Thibault Martin, Partnerships Lead at Dust and Cathy Chen, Head of Product at Wise Business, to answer exactly those questions.
We've summarised the key takeaways from their conversation below.

| Meet the panellists | |
| Thibault Martin Partnerships Lead at Dust | Thibault Martin leads Ecosystem at Dust, where he built the partner program from scratch to help companies deploy AI agents that actually get adopted. Before Dust, he was in charge of tech operations at Doctolib where he scaled technology teams and improved product productivity, and earlier worked in data roles at Lyft. His approach now is simple – empower people to build AI for themselves and make it work for them, rather than having it imposed on them. |
| Cathy Chen Head of Product at Wise Business | Cathy Chen leads the team developing multi-currency payment processing solutions for Wise Business, specifically for businesses selling in multiple markets. As an owner of several small businesses, Cathy is a passionate believer of global payments being fair, fast, and transparent. She draws on a wealth of international work experience spanning across industries and countries – from building Search experiences at Google in Tokyo, to startups in San Francisco, and now to global payments in London. |
Building a hybrid workforce
A testament to AI’s promise to make scaling businesses faster, easier and more affordable, a whopping 88% of businesses say they’re already using AI in at least one business function. As founders scale their teams, however, they’re trying to find the right balance in their workforce. What roles can be managed by humans and which can be managed by AI agents long-term?
Thibault kicked things off by noting that there's no single right answer, “There's nobody who really knows what the ideal blend is going to look like. It's more so about incorporating more AI into your existing processes and going from there,” he explained. For some startups, he noted, that could mean “80% of spend will be in AI and maybe that's effective. Other times it could just be 5% — it all depends on the business’ needs.”

For founders launching today, Thibault sees an advantage over incumbents. “Many companies are retrofitting and trying to understand how they can drive AI adoption at scale.” Founders building AI-native from day one get to skip that step — and it’s changing how they hire. At Dust, Thibault has experienced this first-hand, “the types of people I consider hiring are very different. I want people who can make the most out of the systems that have been built with AI.”
Businesses in highly-regulated industries, like Wise, may have more to consider when onboarding AI agents. Cathy noted that in her team’s own risk review process, AI can handle a large portion of initial business review but human intervention is still essential. “AI is not something we would feel comfortable relying on yet to do a full review.” Still, this is enough to enable the sales teams to partially pre-qualify leads before a full risk review, improving productivity and shortening customer waiting times.
Most businesses will need a combination of hardworking AI agents and human employees. As Hannah put it, combining both AI with humans “is not just about improving efficiency, it's also a better customer experience." While consumers today do expect speed, they also expect a human touch — at least for now.

Unblocking AI adoption
Despite data showing that productivity growth is 40% higher in industries with the highest AI exposure, getting teams to fully understand and implement the technology can be a challenge. Both Cathy and Thibault agreed that one of the biggest points of friction of AI adoption isn’t finding the right models, it’s getting people to work effectively with AI tools.
“The blocker to AI is much more human than it is technological,” Thibault said. Regardless of how quickly organisations onboard AI, it may take a while for people to become comfortable with the new processes it necessitates.
“If you froze the technology right now and empowered everybody to really use it to the top level, you'd already have a radically different organisation,” he added. But collaborating with agents can mean becoming a different kind of employee or manager — one who "doesn't fully own the execution" anymore, and instead has to delegate. Thibault noted that this process is “almost like becoming a manager, but instead of managing humans, you manage digital agents.” In other words, it requires the development of an entirely new set of skills.
For Cathy, the main blocker to widespread AI adoption is concerns over output. She notes, “with AI, garbage in is garbage out.” Despite her slower approach to adoption, the technology is already radically changing her day-to-day. As AI expedites tasks like coding, her job is increasingly focused on figuring out the foundation of the problem she’s trying to solve. “Every hour you spend figuring out what you want to achieve saves you 5 hours down the line trying to get back on track.” While full integration may take a while longer, AI is helping Cathy and her team solve the core problems they need to help customers with.

While both panellists shared an understanding of AI’s ability to improve their work, they also noted that AI isn’t always necessary. In a customer support scenario, Thibault shared, automating a response might look efficient, but if a customer's real need is to feel heard rather than answered quickly, automation is solving the wrong problem entirely. “If your underlying problem is building customer satisfaction, sometimes that doesn’t come in the form of fast answers. I think it’s really crucial to ask the right questions.”
Managing AI spend
In recent months, the AI industry has been hit with headlines of staggering AI bills after a period of testing, learning and ‘tokenmaxxing’. As many businesses now reassess their subscriptions and put budgets in place to control their spend, Hannah asked the panellists: what’s the best approach to creating a sustainable AI system?
For Thibault, “AI is a different kind of spend and you need to treat it as a strategic resource.” He suggested an analogue to AI could be energy, “When you pick your source of energy, you want to make sure that you have some control over it. You want to have resilience and you want to have choice. I think right now, many companies are going for single provider AI solutions and building their entire systems on top of it. The problem is if this provider raises prices or if this provider doesn't have enough computational power to serve your model, you're going to get stuck.”
He recommends that “whatever solution you build your infrastructure on should give you some visibility into how that spend is being made, because when it starts hitting like the ceiling and going way up, you want to make sure that you have ways to control for that.”

For Cathy, managing AI spend can also mean finding more affordable solutions. While frontier models may be trendy, they aren’t always necessary. In fact, most day-to-day tasks probably don’t require a frontier model at all. “If you're just trying to automate tasks, any open-source model should do a good job.”
To determine where you should save or splurge, Thibault suggested blind testing — swapping the model powering an agent and asking a team which output they preferred. "If the difference is not super visible, or it's just a few percentage points, you probably go for the 10x cheaper option."
Fundraising in the age of AI
Last year, AI firms captured 61% of all venture capital investment, doubling its 2022 share. Simultaneously, investors’ expectations have rapidly changed. While it may have once taken months or years for a founder to build a product, AI has accelerated timelines — and investor expectations.
Dust recently closed a $40 million Series B round led by Abstract and saw a stark contrast between its Series A and Series B stages. "Series A, you say okay, we've figured out some kind of product-market fit. Series B is about accelerating," Thibault explained. Seed-stage investors buy into the idea and the team but by Series B, "you need to have some conviction in the value you're bringing to customers and how to convey that value."
The metrics for AI startups are changing too. "There are two types of ways of looking at margins," Thibault said. "If your margins are too high, it means people don't use your AI features. If they're too low, it means maybe they're using it a lot — which could be good, or it could mean your business is going to go bankrupt."
Cathy pointed out how much the landscape has shifted for solo and non-technical founders too, citing a friend who raised $2.5M within two weeks of starting her fundraise off the back of a strong problem and a clear AI spend plan. "In the past, even a couple of years ago, if you were a first-time founder going to pitch VCs without a technical co-founder, they'd be like, 'get out of here.' That's just not the case anymore."
But Thibaut was clear that the bar for quality has risen simultaneously, “Before, you could sell a story or a deck of what this could look like if you got the resources to build. But now that you have the resources to build, the expectation is that you can show a bit more of that — a prototype, something tested on the ground."
Throughout the panel, it was clear that both Thibault and Cathy kept returning to the same central belief: the businesses that will win in the age of AI are not necessarily those that maximise their usage of their technology but those who are strategic about their use of AI and disciplined in their commitment to solving customers’ problems.

Ready to take your startup to new places?
Whether you’ve just launched your business or are already scaling internationally, Wise Business has all the tools you need to operate in 150+ countries. That means one less thing to worry about while you're fundraising or accelerating your growth.
Receive payments from investors overseas, extend your runway with Wise Interest, create multi-currency debit cards for your team and more — all from one easy to use account.

Create your account
*Please see terms of use and product availability for your region or visit Wise fees and pricing for the most up to date pricing and fee information.
This publication is provided for general information purposes and does not constitute legal, tax or other professional advice from Wise Payments Limited or its subsidiaries and its affiliates, and it is not intended as a substitute for obtaining advice from a financial advisor or any other professional.
We make no representations, warranties or guarantees, whether expressed or implied, that the content in the publication is accurate, complete or up to date.