“Ask any programmers about writing a test,” says Diffblue’s founder Professor Daniel Kroening, “you’ll see a look of disgust.”
That’s about the closest the laconic Kroening gets to being emotional. Even for someone who has just led a company to £17m Series A funding round. Diffblue at its essence is an artificial intelligence company, spun out from research done when Kroening, a computer scientist, was a researcher at the University of Oxford.
More specifically, Diffblue is in the business of automating coding. “What we have is a piece of code that generates more code,” explains Kroening, reducing the complexity to as simple a level as he can.
We all have a background in lecturing and computer science if we aren’t able to train up the people with the skills they need then who can?
The initial product application - what Diffblue is selling right now - is centred on quality assurance. That is, the hated test Kroening refers to. The Diffblue product automatically spots bugs and writes tests. Historically this has been one of the most laborious parts of software development (but also vital, especially in this era where software presides over so much of our lives).
The broader vision, where Diffblue is headed, says Kroening, is creating a product that can write whole computer programmes after being given a few examples. “The ultimate goal is to put programming into the hands of non-experts,” says Kroening. “We won’t do it like programmers are doing now, typing into a keyboard. A non-expert can show a computer what you want it to do, and it will extrapolate from this example.”
This sounds altogether high-tech (probably because it is). But Kroening finds himself in new territory: until recently Diffblue’s staff consisted entirely of programmers. “The one non-technical employee was our office administrator,” says Kroening.
But now, as the company aims to begin moving product, the army of programmers will be joined by a sales and marketing team. Diffblue, currently with 50 employees, wants to swell to a 100 headcount by the end of the year.
Kroening wants to create a product that can write whole computer programmes.
Besides an incipient expansion into the US, staffing is where most of the investment will go. “We will grow our development team somewhat,” he says, “but the majority of the spend will be on a sales team and marketing”.
It’s been a learning experience, Kroening admits, but he has received help from his investors, they being Goldman Sachs Principal Strategic Investments, alongside Oxford Sciences Innovations (OSI), and the Oxford Technology and Innovations Fund (OTIF).
The company has added a sales director, John Wood, with a long IT pedigree. It’s an expertise needed to effectively sell and market Diffblue’s complex products. But curiously, this emphasis on pedigree doesn’t necessarily extend to the programmers.
Kroening explains: “AI is a hyper-competitive market. Anyone with even modest credentials has been hired already. So we hire people who are generally good programmers and we arm them with the skills they need.
“We all have a background in lecturing and computer science if we aren’t able to train up the people with the skills they need then who can?” The tactic of hire-and-train also has another unique benefit: it insulates the business from external pressures like heightened immigration control post-Brexit.
This hiring process will endure on the technical side, says Kroening. Diffblue, as it adds new layers of hierarchy, will move to creating two parallel streams of seniority: technical seniority and managerial seniority. “The technical seniority will remain crucial, of course, but we want to give managers the power to lead, too”.
Francois is the deputy editor of BusinessZone and UK Business Forums.