Vasco Pedro, formerly of Unbabel, launches Spinnable. “Business success shouldn’t depend on the number of people you can hire”

For more than ten years, he was the “face” of Unbabel, a company he sold to the American company TransPerfect. Defining himself as a “builder”, Vasco Pedro returned to Portugal and is launching his new startup this week: Spinnable. A platform for creating artificial intelligence (AI) agents, where entrepreneurs can recruit future employees… AI.

“Large companies already have a lot of resources allocated to artificial intelligence, but small companies are lagging behind because it is difficult for a company with 5 or 6 people. In a home renovation business, they are focused on their business and don’t have many resources to hire artificial intelligence people to create agents, so they end up losing competitiveness”, says Vasco Pedro.

“We are able to help scale businesses that are currently small but don’t need to be small; it’s a matter of a lack of people. What we say is ‘business success shouldn’t depend on the number of people you can hire'”, he affirms.

The start-up, which wants to change the way people work with AI, is starting with 15 employees, 10 of whom are “non-human”. Among the humans are Gil Coelho (former head of product at Unbabel) and Fábio Kepler (former director of AI research at Unbabel). The promise of making conversations between humans and technology more humanised has already resulted in the raising of two million euros in a pre-seed round led by HeartCore, Florent.vc and Alphagraph.

Spinnable is starting with a waiting list of users, but the goal is to be fully on the market within a year. “Our goal is to keep the team small and continue on our path. The goal is profitability”, says Vasco Pedro.

After selling Unbabel, when asked about the future, you said: I am a “builder”. What are you building now?

What I am building is an artificial intelligence agent platform [Spinnable] that allows companies that do not yet have many resources to scale their teams. The idea is to scale beyond the human limits of a company. It’s a bit like a continuation of artificial intelligence working with humans, but now it has nothing to do with translation. Like us, by enabling companies to be more scalable and more competitive, we [manage] to give humans increasingly human experiences. So, basically, instead of constantly having to learn new tools, we are working with humans who are part of our team and with AI agents, also part of our team, who help us to scale the company.

You also use the tools you are creating. Dana Allen, your assistant, for example, is an AI agent.

Immensely, immensely. We are the biggest consumers of our own tool. In our team, at the moment, there are five humans and ten non-humans.

The discourse surrounding AI oscillates between euphoria over productivity gains and fears that it will lead to job losses and make it more difficult for young people to enter the world of work. Does your solution fuel that fear, or does it solve it? What can you add to the discussion?

Right now, there is an opportunity for political leaders to understand what the new social contract should look like. What do I mean by this? I think there will be some very positive developments. For example, a company that uses more artificial intelligence can achieve higher productivity per employee and offer each of its employees better conditions. I see this at Spinnable, in new AI companies. It is also increasingly easy to start a business — you can go to Spinnable and say, “look, there are two of us, we don’t know anything about marketing, products, or anything else”, but you can hire at Spinnable and launch a company without having to hire I don’t know how many people — this increases the business fabric and productivity. But there will also be challenges.

If the companies of the future are going to have fewer people — which will happen, because it’s very difficult to manage a company with two thousand people, there’s a lot of wasted energy in getting people aligned… — if, on the one hand, there are more companies, but on the other hand each company has fewer people and hires fewer people, how is that going to work in practice?

There is huge investment in artificial intelligence, which, at the moment, genuinely has both positive and negative potential, and there needs to be political leadership to understand how we, as a society, are going to absorb and use this in the best possible way.

This dialogue is just beginning, it has to continue. I think Spinnable will be part of this dialogue, we have to be part of it, accelerate the need for this dialogue and try to constructively help it reach a positive conclusion.

It is easy to create a utopian image of where we will be in 10 years’ time, when we will all be doing very well. The problem is the transition. I fully understand that when people hear “I’m going to hire an artificial intelligence agent and I probably won’t hire anyone else”, their reaction is “how am I going to remain relevant?”.

Until the industrial revolution, age brought wisdom and value. If you had a problem with your field, a drought, you would go to the village elder, who had seen more, and who would tell you how to solve it. With the industrial revolution, each generation knows more about technology than the previous one, has an advantage, and the acceleration of this.

What I’m seeing with Spinnable is a bit the opposite. What we’re doing is saying, “the world was built for human beings, so why don’t we make AI work in a human way?” By doing this, we are democratising access to artificial intelligence. For example, one of the agents I created was for my mother. My mother is in a retirement home and, like most people her age, she doesn’t have a huge technological capacity, but I created an agent she talks to on WhatsApp. It’s been super useful, she feels super accompanied.

This ability for everyone to use artificial intelligence and not just young people having access to technology, suddenly there’s an advantage, again, of experience. That’s something interesting that I see happening.

An executive at a large company hardly uses any tools anymore; all they do is talk to people who report to them. That experience will be the human experience in the future. We will spend more time talking to people, whether they are human or the human experience but with artificial intelligence, but the experience will be more human. There will be fewer tools, logins, interfaces, learning new things, and more understanding how I continue to talk and interact in the easiest and most natural way for me, which is through conversation.

In the case of LLMs, such as ChatGPT, there is often talk of model “hallucinations”. When selling AI agents to companies, what safeguards are you putting in place to prevent these “hallucinations”, which require excessive verification?

There are three important components. One is that when humans hire an intern, they also make mistakes; they need to learn. The same is true with AI right now. When you hire an agent, think of it as an intern who needs a period of training, during which they may need more feedback and supervision.

The other is through the permissions you give them. Our agents are designed to be autonomous team members. When I give ChatGPT access to my Gmail, it can do everything I do. But if I give a different Gmail account to one of Spinnable’s agents, I can put restrictions on that account itself, on the software it has access to, which helps.

Then there’s the guardrails part, which is also about not putting the agent in positions of too much responsibility and assuming that it’s early days, so we have to be careful. So we’re going to launch and do this gradually. That means we’ll have a waiting list, work with users, be attentive, and understand how the agents are behaving. We tell people, for example, not to give an agent their credit card yet.

Users themselves need to realise that this is a super powerful technology that can help immensely, but we also have to be responsible in how we use it and understand its limitations. So it’s a two-way street. On our side, having a series of guardrails that allow us to control the agents, and this is constantly improving, will become less and less of an issue, but people also have to use it carefully.

Is there a specific sector you are targeting, or is it cross-cutting in its applicability? You mentioned your mother’s case…

From my mother to Airbnb companies… In fact, what we’ve noticed is that the less technical it is, the more people appreciate it because instead of having something complicated to use AI, they have a metaphor they understand: “I may not know how to do complex workflows, but I know how to talk to a person, I know how to say what I want”. You are the manager of these agents, you have control, the ability to specify, but in a human way. We are not seeing any limitations in areas at the moment, we are seeing a very horizontal adoption, we are more focused on SMBs [Small and Medium-sized Businesses, micro and small businesses].

Large companies already have a lot of resources allocated to artificial intelligence, but small companies are lagging behind because it is difficult for a company with 5 or 6 people. In a home renovation business, they are focused on their business, they don’t have a lot of resources to hire artificial intelligence people to create agents, so they end up losing competitiveness.

We are able to help scale businesses that are currently small but don’t need to be small; it’s a matter of a lack of people. What we say is that the success of a business should not depend on the number of people you can hire.

Are you bootstrapping or are you in the process of raising capital?

We have already raised capital. Initially, my plan was not to raise capital, but the market was super, super, super active. In fact, I posted on LinkedIn that I was doing something new, then ECO published the news and that created even more appetite and we have already raised capital. Our goal is to keep the team small and continue on our path. The goal is profitability. That is, to make a product that truly brings value and that people are willing to pay for. That is the best measure of whether this brings value to the world.

Who are the investors and how much capital was raised?

We have some angels, long-standing investors, a few other people, and we have a main investor, HeartCore, which is also an investor in Luca [edtech], if I’m not mistaken, they have a number of unicorns. I really liked Max [Niederhofer, partner in the European fund] and they invested €2 million. That’s enough for us.

Investment in AI is on the rise, there are many unicorns being born, a lot of investment being made in the sector, but recently the stock markets have been in the red with fears of a crash, of an AI bubble. How do you view this?

That’s why we decided to raise some capital, because it’s good to have a cushion. I’ve seen this at Unbabel, ups and downs, and it’s good to have capacity. Ideally, you don’t need money. I have a maxim at the moment, which is “product market fit is the only thing that matters, everything else is kidzania”.

At Unbabel, we raised a lot of capital, which is one way to go. There is a time to raise capital, when you actually have product market fit and you are scaling. We are currently in the exploration phase. We are exploring how a platform that has a human metaphor fits into companies and interacts with other human beings. One thing we have seen, which has been fascinating to me, is that our users, who are tech-savvy and male, have more difficulty with the metaphor. For some reason, when our users are women, they connect much more with this metaphor.

The democratisation of AI at this level, in terms of gender and age, is the path I want to explore: how social intelligence can fit into human society, in a human way, that makes our experiences more human and more capable.

Do you still live in the US? With your work logic, I imagine it’s easy to maintain a remote working model.

I’m in Lisbon. I’m at Atrium Saldanha right now. This is where our office is. Actually, Spinable isn’t remote. That’s another decision we made. The advantage of having a small team is being able to work physically with other people. I have nothing against remote working, obviously there are very successful companies, but my culture, what I like, is working physically with other people.

Sword Health, Unbabel’s partner in the PRR AI consortium, has announced an investment of €250 million and the ambition to position the country as an AI hub in the health sector. When countries such as the United Kingdom, Germany, or France are investing heavily in AI, just to mention a few examples, does Portugal really have that capacity, or is it more of a wish?

I think we have the talent and human resources. There has been an influx in recent years, and Portugal is currently on the rise in terms of a place to build and live. That may change. There are a number of policies — for example, the nationality law has now been passed — and some restructuring of some of the incentives that were being used to attract people. But it is true that these incentives have existed for many years.

I know Lisbon better than Porto, but in Lisbon I see a fantastic culture of entrepreneurship. There are a lot of people living here. For example, one of the funds that ended up investing in this round was Florent VC, one of whose partners has been living in Lisbon for six or seven years. The culture is changing, and this is attracting very interesting people who, together, in turn, also create a positive cycle, in which local people are exposed to this without having to go abroad.

So, I still believe what I believed before, which is: if we can create a nucleus in Portugal of interesting companies doing interesting things, people will want to stay because we can pay them, we can retain them. Now, we have the potential, yes, but it’s not a guarantee. All these initiatives are still at a relatively fragile stage. There are many other countries that have an advantage. If you think about Israel, it has more unicorns than the whole of Europe, not because they have more people, but because it takes time and effort to invest.

I have no doubt that we have the capacity to do so. I have always been a huge advocate for Portugal. We are experiencing a good moment, but I think there is still a lot to do and, of course, nothing is guaranteed. A few years ago, people were saying, ‘Oh, San Francisco is finished, it’s dead.’ It’s completely on the rise with the AI revolution that’s happening, so cities also change and reinvent themselves. But the energy I’m feeling in Portugal and Lisbon, which is what I know best, is fantastic. It’s one of the reasons why I wanted to build something else here.

Another Web Summit kicks off today. Ten years on, does it still make sense for the country? Do people come for the “postcard” and then find that the reality is different?

The work is not done, so I think the Web Summit brings a lot of value to Portugal and Lisbon. I’m going to speak there this week, as I have done since I came to Lisbon, and I think it brings immense value. The discrepancy between the Web Summit and reality is smaller now than it was 10 years ago. You have other events, Sword Health is creating an AI hub for Health, we have Unicorn Factory holding events, the IA Hub in Alvalade bringing together a number of start-ups. The delta is smaller. As long as we can keep the Web Summit in Portugal, I think it brings immense value.

It will remain in the country until 2028. Is this a negotiation the government should pursue?

Yes, but, as in any negotiation, not at any cost. But I think it definitely adds value. In Dublin, when the Web Summit left, they created the Dublin Tech Summit, and I think Dublin lost a lot in terms of relevance.

There was a combination of things in Lisbon, I don’t think they were specifically planned, but they coincided and they are all helping: people can get here easily, they have references, Portugal is a great country to live in, there are many interesting opportunities, many advantages to being here. We are aligned with London, everyone speaks English, we are closer to the United States.

It’s a country that, in general, has good relations with everyone in the world, where, for now, and hopefully forever, people who move here feel welcome, embraced, and want to make their life here, they like to bring their family. Many of my friends who moved to Portugal are people who bring immense value to the country.

We only see the less positive side of immigration in the news, the influx and all the controversy surrounding it; I see more of the other side: a number of people who have built incredible things in their lives, who have decided to move to Portugal with their families, who are contributing, who want to create things, who bring their knowledge and resources, and that is super positive.

You’re launching the company this week. What are your ambitions?

Profitability. A maniacal focus on having a product that brings value. I think that in the next two or three months, we’ll have a waiting list and start letting people in. I would say that by the end of the year, we will be available to everyone, and in a year’s time, we will have a significant number of companies that are able to scale more because they use Spinnable agents. That’s the goal. When we feel that we genuinely have product market fit, then yes, raise serious capital and scale globally.

After the sale of Unbabel, you stayed on to hand over to TransPerfect. I admit I expected it to take a little longer. Did it go well, did you leave the company in good hands?

TransPerfect is a company with 12,000 people globally, it has very established processes, it buys a lot of companies. I’m here, I’m still here to help with whatever is needed, but I think my role right now, my usefulness, my value is in creation.