Supervising Disruption.

Supervising Disruption? Some kind of clever ML pun? No – there is a trope normally attributed to Marc Andreesen that goes something like this. If it’s a consumer play, back young people doing something you don’t understand. If its B2B, back founders from the industry who have done it before, preferably together. It seems to me this next generation of deep or frontier businesses are about doing both – you need fresh eyes and brains to ask disruptive questions, but you still need experience to navigate distribution and enterprise priorities. How well prepared is the UK/EU venture environment for backing these kinds of projects?

Not very is my experience.

When we first started investing in deep-tech companies 5 years or so ago, there were few folks out their very interested. The ecosystem has been transformed – more specialist funds, more specialists – and just more funds – chasing everything that has AI, Bioinformatics, drones or robots in the business plan. More talent, more experience, and more money has had an impact on valuations and the ambitions of founders, which is in many ways good. But I’m not sure that the rest of the ecosystem has caught up. I’m not sure there are enough later stage investors in Europe who will back technology and vision but little commercial proof with big bucks – although I am sure there are lots of non-specialists investors with one or more deep tech deals on their books that will get frustrated at the two year mark when they are not seeing that magic £1m ARR they’re told to expect on the horizon.

This blog is the first of three that are my attempt to share some of the best practise and lessons learned from our experience. It’s a work in progress, stuff rattling around my head for the last few months, some parts are slightly contradictory, and some may reflect our own naivety. I mostly use AI examples to illustrate points, but I think most of the comments are broadly applicable.

Setting the Scene

Another week, another demo day, another group of well-rehearsed entrepreneurs pitching their world changing ideas. This particular week’s group was mostly fashionable frontier tech start-ups – two PhDs and an MBA taking something based around AI and disrupting medicine, healthcare, finance or insurance.

It reminded me of that old joke about the 2 micro-biology PhDs who walk into a bar.

“I haven’t seen you two in here before, what brings you to town?” says the barman,

“We’re researching the impact of coprophagial digestive enzymes on microbiota”

“Wow – that sounds hard! How can I help?”

“Two pints of cloudy cider, some pork scratchings and one of those hard-boiled eggs should do it” comes the reply.

Ok so that isn’t an old joke, I made it up, and you’ll forgive me a small dose of occupation-appropriate cynicism. In reality no one loves seeing science fiction turned into reality more than I do, and trying to understand the state of that science with smart impassioned founders is one of the true pleasures of this job. However, as I sat in the audience I couldn’t help but wonder how many investors really understood the journey that they were about to go on. The pitches give the impression of imminent world domination – the reality is most teams are years away from product-market fit and even longer from substantial revenues.

Landmines Along the Journey

I want to start with calling out some of the common mistakes we’ve seen or have been guilty of. If AI is all about pattern recognition then guilty as charged – here are some landmines to identify and avoid. I am sure there are many more, but these will at least make you sound more informed at board meetings.

• Confusing R with D – many engineering problems in frontier-tech are still in the realm of pure research. No one has ever actually solved them. Here’s the scenario – “Hypothetically such-and-such an approach should give good results – but we won’t know until we’ve spent 3 months playing around with it. And if we’re wrong, we need a new hypothesis”. Hard to plan and budget for – make sure you understand what is R and what is D in your engineering plan.
• Confusing theoretical results with world domination – producing impressive results on small or standard datasets, or from a limited pilot is useful. But it’s not problem solved or time to declare world hunger sorted. Limit cases, strange false positives or negatives and generally unanticipated outcomes from customers only show up in the wild – and only when you are processing large numbers. BTW, included in that may be the discovery that the team’s back end deployment skills are not as robust as their front-end science skills.
• Confusing technology with product – big one. Selling product into enterprises means there is a ton of tech debt you are going to need to put up with that isn’t sexy, innovative or differentiating. Authentication, user groups, dashboards, security, roll-backs, back-ups. Enterprise features you absolutely need to nail.
• Confusing charisma with use cases – the PMF journey is long and arduous. There are founders who can sell anything to anyone, and there are certain enterprises who will buy anything from anyone charismatic. Squashing your triangle shaped technology into a square shaped customer problem can be a good learning exercise, but it should not be confused with PMF. Clear use cases are replicable, scale, solve real problems, have real ROI – and can be sold without the charismatic founder.
• Confusing Innovation Teams with Customers… – it sounds great! “really cool stuff like yours is absolutely on our roadmap. Join our incubator, have a £25k pilot – and come and make me look good in front of my boss!” Corporations want to be seen to be innovative, partly of course because innovation is generally a good idea, but in most cases because it’s good marketing, good for hiring or makes senior execs feel young again. Corporate innovation programs need fodder – that fodder is shiny start-ups. Do not confuse these good folks with people who have real budgets, control access to procurement or are responsible for actual deployment of IT. One global innovation lead at a huge corporate told me they over the last few years they have worked with 1000s of start-ups, done pilots with several 100, but only a handful have reached production and real revenues. That sounds about right.
• …and just confusing customers – when you are in the technology supply side of the business you seldom think about the demand side. Think about how many different applications you can deal with on your mobile. Now think about that translated to the enterprise – customer mindshare is a real and growing issue. Very few problems don’t already have some kind of solution, people are overwhelmed with new technologies, and probably still struggling to grasp the details of the last IT roll-out. Unless you are offering a 10x improvement it’s going to be super hard to get anyone’s attention – particularly true in my experience for cyber, marketing or customer care type solutions.

And I’m sure there are more – these are some we’ve seen more than once. The good news about repeated mistakes – that which does not destroy us makes us stronger – is you can turn them around into a series of observations about what you should do differently next time – and hopefully identify progress milestones. That is the subject of next week’s blog.