Team building for startups Entrepreneur First pushes people to partner up and encourages them to base their search for an idea with potential on their special expertise and experience – which flies in the face of the Centre for Entrepreneurship’s suggestion (which I decried last week) that universities should simply run courses in entrepreneurship for graduates.

Many of the well-known startups have begun with a couple of people happening together upon an idea. The details of these meetings differ, but the story is the same: a tech-fuelled version of romantic love. Apple, Google and Airbnb all began that way. So in Silicon Valley – and across the startup universe – it is assumed this is the only way to do it. “There’s this assumption that the easy parts are building a team and finding an idea, so the accelerators say ‘Come back to me when you’ve got those bits’,” Alice Bentinck, COO of Entrepreneur First says. “Why do we leave this to chance? Why can’t we help people learn what is a good team?”

EF is an accelerator for individuals, which since it opened its doors in 2011 has launched 143 companies with a collective value of more than £400 million. New recruits come in, meet a co-founder, develop an idea and build a startup from scratch. If YCombinator is getting together with friends, EF is speed dating and Tinder. “People want entrepreneurship to be romantic,” Matt Clifford, CEO, says. “We’re trying to find a way for people rapidly to speed up and increase their likelihood of finding love.” Speed is critical – because, at EF, participants have to pair up in just three months; spending time getting to know people isn’t an option. Anyone who can’t find a partner and a commercially feasible idea by the halfway point of the six-month process is asked to leave immediately.

Step 1 Partner up. Only 16 per cent of VC-backed startups have one founder, so if you want venture backing you’ll need a co-founder.

Step 2 Don’t commit too soon. Co-founding isn’t like marriage. The wrong partner can sink a startup – so if it’s not going to work, leave before you get stuck.

Few people thought that Clifford and Bentinck would succeed; investors would say: “‘There’s no way you could build teams from scratch.” But so far it appears to work. The first definitive piece of evidence arrived on June 20, 2016, when Twitter announced the acquisition of EF graduate Magic Pony Technology for a reported $150 million (£120m). Magic Pony’s founders, Rob Bishop and Zehan Wang, both studied together at Imperial College. Even when they arrived, they didn’t start working together until eight weeks into the programme. Then Bishop, one of the first engineers at Raspberry Pi, suggested using Wang’s artificial intelligence PhD work to speed up image processing. Three weeks later, they had a prototype; eighteen months later, they were tech millionaires. So too were Bentinck and Clifford. EF’s eight per cent share of Magic Pony, purchased for $16,000, was now worth $6.5 million.

To its founders, EF’s success suggests a revolutionary conclusion: entrepreneurs, long presumed to be born, can in fact be made. “It’s basically saying we no longer have to wait organically for these guys to meet at Harvard or Stanford; you can actually take that process and do it at scale,” Clifford says.

The first instruction from EF is simple: get into a pair by the end of the week. “Everyone hates it,” says Bentinck. “But you can’t understand whether someone’s good to work with until you’ve worked together.” There’s another reason: by tracking teams on previous courses, EF observed that about half of successful teams form within the first two weeks.

To help participants get together, EF tells them to focus on their “edge” – their strongest point. Over ultra-competitive board games and cooking challenges at a pre-weekend, the group were given advice on how to meet people. First tip: pitch yourself, not your project. “What I learned from the talks was, if you come in with a fully fledged idea, it’s quite difficult to find a co-founder, because they feel like you own the idea,” says Johnnie Ball, a former trader who left a job at an energy startup to go to EF.

Focusing on what you know may sound obvious, but it runs counter to the dominant school of startup ideation: solve a problem you’ve experienced. EF turns that process on its head. Rather than thinking of problems, it advises, start with what you know, then go in search of ways to apply it.

EF is a business, not a research institute, so to stay in the programme the teams have to build things people actually want. To make sure they’re moving in the right direction, the teams need to locate, contact and, eventually, sell to customers. “We say to them all the time, go talk to your customers,” Clifford says. More broadly, they need to become businesspeople. This is EF’s bet: that it can teach technical founders to think about commerce. That’s one purpose of the weekly pitches. It forces the group to learn how to sell. (Every Friday at 11am, each team presents their work to the rest of the group.) This is their chance to make comparisons, impress EF and show off to potential partners.

Step 3 Find your edge. No matter how tempting it is to expand your horizons, it’s what you already know that’s going to give you a competitive advantage.

Step 4 Stay in school. EF started out taking graduates, but found that they weren’t experienced enough. If you’re still in school, you might want to stay there.

 To its critics, Entrepreneur First is little more than a glorified meetup. “They put people in a room and that’s it,” says Nathan Benaich, a partner at Playfair Capital. “There are many ways that young entrepreneurs can get the benefits of the process without giving away a relatively large chunk of long-term equity for limited short-term value.” It’s a familiar complaint: EF is being accused of not having an edge.

Their first programme began on September 1, 2012 – in spartan surroundings. More significantly, the focus was unclear: there was a mix of technical and non-technical founders, and no mention of concepts such as edge. But, to everyone’s surprise, it produced 11 companies, including four that eventually sold (although EF didn’t raise a fund until the second group, so it didn’t make any money from the deals). Delighted, Clifford and Bentinck forged ahead with a second programme. They felt they had startup-building cracked, but things soon started to go wrong. First to fall apart was the team-building process.

At the beginning, Clifford and Bentinck believed that founders would come together and stay attached. Over the summer of 2012, they organised a series of hackathons to help participants decide on their final partners. Then the course started – and, one by one, the pairs started breaking up. The situation came to a head at with a mass breakup at the end of October. They called it ‘the Halloween massacre’. “This was a crisis moment,” Bentinck says. “We weren’t sure how to fix it. And then people suddenly started saying, ‘Oh, well, why don’t we work on that idea together?’ That kept on happening. We were like, ‘Hang on a minute, maybe there’s something in it.'” As new pairs kept forming, a method was born: instead of pushing co-founders together, EF would pull them apart.

One way to do this is to help people with their breakups. If a team is struggling and won’t break up, EF will step in and do it for them. One participant found this out on the Friday of week six, when she and her partner were approached and told that they could not work on this any more and had to split.

Three months is a long time at EF. Of the 100 people who start, 30 fail to make it to the halfway point. Some never find a partner; some do but can’t make it stick; others drop out for different reasons entirely, such as the AI entrepreneur who leaves because he’s been given €50,000 (£42,000) by Google to build an automated fact-checker for fake news. But in the final weeks there is a last-minute rush to form partnerships, so 35 teams present themselves for consideration. Some succeed; some don’t.

Like its startups, EF’s process works less well if you don’t have an edge. Someone may be a skilled coder with a bachelors degree in graphic design and another in computer science, but among the PhD graduates and data scientists of EF, he or she is a jack-of-all-trades coder. That would be fine if she was content to take the lead from someone else for she has the traditional entrepreneurial qualities of drive, ambition and an appetite for risk. But at EF, these can sometimes be a hindrance.

The same difficulty occurred in 2011, when the second group formed new teams. As they came together, one type of person was excluded: with all but one exception, the businesspeople dropped out, and the startups were created by founders with technical backgrounds. Yet this failure, too, suggested a solution.

Step 5 Test your idea. If you want it to work in the real world, ask customers what they think. Then ask them again.

Step 6 Pivot – but not too much. If things go wrong, make a change, but build on your experience and contacts.

 When the second course ended, Clifford and Bentinck decided to change their approach. From now on, EF would ignore businesspeople and look exclusively for technical founders. The decision was controversial. When Clifford and Bentinck announced it in 2014 at its third Demo Day – the showcase graduation event held at the end of each course – Bentinck remembers the assembled investors gasping in shock.

But whether consciously or not, Clifford and Bentinck had timed their move to perfection. Technology is shifting away from general software and towards mathematical algorithms. In this world, business savvy is no match for a PhD in computer science. That same year, Google bought London AI startup DeepMind for £400 million, creating an instant pool of local machine-learning millionaires. Among investors and entrepreneurs alike, AI is in hot demand. And EF – the creator of Magic Pony, now taking applications from one in three Cambridge computer science graduates – is the place to find it.

EF is growing geographically and financially. In September 2016, it launched its first international branch, a 100-strong programme in Singapore. The same month, it announced that it would be funding companies for two years, thanks to a new £40 million fund run by Moonfruit co-founders Joe White and Wendy Tan White. Eventually, Clifford hopes, EF could even displace YC. “What they’ve managed to achieve is fantastic and inspiring,” he says. “But I fundamentally think that they’re still an old-world institution. They’re basically just investing, and they do so little for their companies.” The question is: can EF can do more?

Clifford and Bentinck believe EF will change the world. “People still don’t get how profoundly radical it is,” Clifford says, “to be able to take people as individuals and turn them, with some probability of success and massive creation of value, into companies. The EF process is certainly not infallible. It’s not the place entrepreneurs are made. It helps certain kinds of people – focused, technical, experienced – build certain types of companies. If it succeeds, it will be because those companies have become much more necessary.

This is an abridged version of an article first published in the May 2017 issue of WIRED magazine.

John Whatmore, July 2017