Image

As great UK AI Scaleups grow – some new challenges surface

At Digital Catapult i have the privilege to meet lots of leading-edge companies every week and, after a while, the things they have in common start to surface and become more noticeable. In the last few months, a few recurring challenges among early stage deep tech AI companies have caught my attention.

One of those challenges is around deep tech companies who want to develop platforms to license their AI solutions. Like many SaaS companies, they find that they often need early foundation clients to help build their place in the market and prove out their concepts and gain those really important early revenues.

Interestingly, AI businesses in particular, which are rooted in deep tech, tend to be quite diverse in the range of verticals they find themselves entering. This is frequently because the solutions their algorithms offer may be equally applied across many domains.

The process that leads a company to find its first customer with a dataset they want to share is often a somewhat arbitrary and very time consuming process. That’s what gets the company kick-started and it is only when a client with data to be trained shows up, that it leads the business into a particular domain specialisation.

So one of the challenges is how to retain focus within one domain or industry sector, rather than being diverted off into various different sectors which don’t complement each other and which, although they might happen to yield data-rich clients, will ultimately make it difficult for a small AI company to scale.

The argument is always made that the early customers of small tech companies will essentially pay to build the product and then the company can roll it out to everyone else and scale that way.

This is a tried and tested method, and a sensible approach to pursue. Inevitably however, trying to find those early clients takes a long time and when you do sign them up, they want the work to be a pilot and to pay heavily discounted fees.

Then, chances are,  the client will swamp the development team with very particular and granular demands about how the integration with their legacy systems will work and what kind of APIs need to be created – in other words lots of questions which are very bespoke to that customer rather than generic and applicable across the board.

So although useful cashflow revenue is generated, the company becomes almost entirely focussed on the needs of one or two early customers at the expense of developing a scaleable solution.

Inevitably, these challenges which are familiar to many platform businesses, get more complicated in the context of AI companies.

One reason for this is because the development team is almost invariably made up of a team of highly motivated, extremely well-qualified, sometimes academic folk, who love getting their brains around gnarly intellectually challenging problems.

Clients, of course, love that they are getting groups of PhDs or post-doc’s focussed on their problems. The challenge for the company which wants to scale up, is that building out the “rinse and repeat” product that will enable it to scale requires an entirely different set of skills in a development team.

 

Very often the platform development team needs to be made up of a totally different group of people and not the group around which the company was originally formed. The skills needed are very different from those of the algorithmic design and training team, and from the data cleaning team and indeed the data science team (who at the beginning of course are all the same people).

The product development team requires strong user experience knowhow, great workflow design, and rigorous organisation of the information architecture. These are generally things that Machine Learning PhDs are less interested in. So in fact, very often, a product development team does not exist at all on the org chart and the company will require further capitalisation in order to be able to hire them.

Few of these considerations figures large in the minds of founding teams, who very often form around eminent academics or groups of graduating doctoral students. Yet, if they had stopped to consider how to scale the high-grade output of their AI dev team, they would recognise that other more established skills at equally challenging levels of accomplishment, may also be required in order to make scaling their output viable.

We have really great AI businesses in the UK, we need to spend more time coming up with the best and the most appropriate support programmes to help them scale.

 

(This blog originally posted on Digital Catapults website https://www.digicatapult.org.uk/news-and-views/blog/understanding-the-recurring-challenges-faced-by-ai-scaleups/ )

Advertisements

Apple acquires Shazam

shazamI first met with Chris Barton and Philippe Ingelbrecht in Berkeley in around 2000 when the idea of Shazam was just beginning to form. The idea that Avery had created such a smart algorithm that it could read a recording from the noisiest of ambient sound contexts was amazing. Today we would call it AI, but then we just thought it was super cool and that everyone would soon be holding up their mobile phones and figuring out what was playing in that pub, bar, cafe, etc. Of course the mobile phones were pretty clunky then and they were much clunkier in the US than they were in Europe. And folk in the UK were using their mobiles much more adventurously and frequently than anyone in the US. Europe was ahead and that was why Phillipe and Chris decided to come to London to set up Shazam. In early 2002, post 9-11, London seemed like a safer place to start a business too maybe. I spent a few months advising them at that time on how to move things forward, who to talk to and where to set up in London.

shazam There was an interesting IP conundrum that never got fully resolved and that was where ownership resided in a database created from the fingerprints of audio recordings.  I remember wondering how long it would take before the major labels would demand a piece of the action,  even though the database that Shazam created was based on a hash of the audio, not the audio itself. It remains a potentially grey area as to whether this database which is derived from the “fingerprinting” of individual songs is an associated work or not. The case has never come into a court because the majors took some equity instead, which seemed reasonable enough at the time, I guess.

The thing that really made Shazam remarkable was that the company name became a verb and changed consumer behaviour. It took ten years and the invention of the iPhone before that could happen. Who would think that you could get a startup to change the way that people behave when they hear a piece of music they have never heard or want to know the name of or buy. When people say that apps can change behaviour overnight they might be right, but it can also still take decades.

It will be interesting to see how quickly Apple can integrate Shazam and turn it to its competitive advantage, but simply directing all queries to a subscription for Apple Music or a purchase of the download from iTunes will drive revenues. The data that Shazam possesses on the geography, time of day and frequency of searches for music will be of enormous added value to Apple too.

Of course, Chris and Philippe are long gone, but Avery and Dirajh stayed the course. From startup to a $1bn valuation and then an acquisition by Apple – that’s quality!

 

We are still Europeans

The digital and creative intelligentsia of Europe has flocked to the UK in recent years. We have attracted folks because the UK seemed like an enlightened, forward thinking, inclusive place to come and work and think and create. Their skills, inventiveness and breadth of outlook have enriched our universities, our startups and our national institutions. As we wake up to this decision this morning, we must make sure,  loud and clear,  that these brilliant people are still and will always be welcome. This is true for all the visiting workers we have hosted here, but for our creative and digital community it’s vitally important. I don’t know of a startup that hasn’t benefitted from a pan-European work force – not just for their skills but for the global outlook that being cosmopolitan creates. When I think of the difficulties we have encountered when we have had to try to secure visas for US citizens wanting to join us and contribute to our economy, it is clear that we must urgently insist on avoiding any changes to residency rules for folk like this. We may have decided as a nation to leave the EU and the economic consequences of this will be uncertain for weeks to come and  are already clearly negative. That said however, we cannot allow our community to retreat into some inward island mentality. The country may have chosen to leave the EU, but that should not prevent us from standing as culturally European and citizens of the world. 

Aside

Check out a few of the amazing people I’ve had the privilege to interview at the Mama event in Paris over the last few years:

Nick Mason    Nick Mason

Redim_SeymourStein     Seymour Stein

Keith Harris  Keith Harris

 

Bring back the music nerds!

Musical evolutions are hard to trace sometimes. There is a section in the opening of a song called the Dark of the Matinee by Franz Ferdinand that, to my cloth ears sounds exactly like the beginning of the immortal Jewish folk song Hava Nagila – it comes right at the beginning of the track and lasts about 22 seconds before repeating again throughout, if you want to check it out. Then there’s the tiniest brief line in the Chilli Peppers’ Zephyr Song (just about at the 2.04 minute mark) that seems to morph weirdly for a moment right into Oliver’s Army by Elvis Costello. Argue with me about that one if you will, but I swear, it’s there somewhere.

 

Sometimes these are just things that occur to you when you’re singing badly in the shower (is there any other way to sing in the shower?) and sometimes they might be deeply buried treasure that even the song writers didn’t really think of. Then at other times they’re totally sophisticated samples planted or extracted (depending on your point of view) right in the middle of a song to make us all feel smart or just to get off on a brilliant musical juxtaposition or cultural reference or even maybe to achieve some subtle commercial goal. A sample is a way of paying respect or making a connection with someone else’s scene. We all delight in spotting these things, they enrich not impoverish the music.

 

Sometimes, it’s a bit of a joke or is it faintly sinister when you are taken all the way back to an embarrassing nursery rhyme – Hey Diddle Diddle, the Cat and the Fiddle and find it reappearing in a way that still makes me cringe in the classic Aerosmith Walk This Way, then even more famously redone by Run DMC and then riffed on by Stiff Little Fingers in No Sleep Til Belfast which was itself a cover of the Beastie Boys’ No Sleep Till Brooklyn, which was then sampled brilliantly in Fat Girl by Easy E of NWA featuring none other than Freshly Done. Well what more can I say? The path of true love never did run true.

 

So what are we to make of the wily ways of musicians, songwriters and producers in their apparently endless reworking of respect and satire, of tribute and rip-off. When exactly do the Blurred Lines get crossed into something where, well, you just Got to Give It Up? Well never, allegedly, if musicologically the songs are technically musically different and it’s just a sound or a vibe that kind of connects them. Except, that is, in the case of Bitter Sweet Symphony, which I have talked about extensively elsewhere in a TEDx talk I gave a couple of years at the British Houses of Parliament. Suffice to say that what the Rolling Stones may have borrowed from Gospel and the Blues was all apparently fine, but when someone else wrote a classical composition based on their borrowings and the Verve borrowed from that – well there was all hell to pay – or at least all the royalties on the track – for ever. Perhaps ABKO Records might like to re-think the artist-relations politics of that particular issue sometime before Richard Ashcroft gets too old. But I’m not writing this to complain about the injustices of sampling or the appropriateness or otherwise of the protections that copyright laws afford. On the contrary, it’s the joy of discovery that interests me the most.

 

When exciting new streaming music services seem to be fighting for differentiation on a daily basis, it’s the journey of music itself that ought to give them the best chance to educate us. WhoSampled.com is one of my favourite but most under-exploited music resources, because when I sign in to my favourite streaming service, whichever one that is, I would love to be educated about the music I’m listening to. I would love to be presented by the Horace Silver Song for My Father that supplied the key riff to the Steely Dan song Ricki Don’t Lose That Number or on hearing their classic Do It Again be led into a brilliant mix with Michael Jackson’s Billie Jean by Club House (Michele Interlandi and Stefano Scalera) to hear how those bass grooves interweave so perfectly.

 

The musical skills that lie behind the perfect playlist are sometimes just the perfect segueway from one track to another, but at times it is the musical connections and real interconnections buried deep inside the songs themselves that make for the richest musical experience. Only the rare music afficianado and blacksploitation fan would probably have gone back to the Chi Lites’ Are You My Woman? (Tell Me So) – but Beyonce made its brass stab the signature of Crazy in Love and now far more people would have a reason to go back and revisit the track. So that one might start to look like strategic back catalogue marketing. Equally, would Major Lazer have quite the same profile if Run The World (Girls) hadn’t featured their utterly identifiable beeping whooping sample from Pon De Floor or was it more that Beyonce was garnering some fresh street cred from the Lazer collective? What comes up from the underground to the overground takes everyone forward.

 

When iTunes first appeared and made digital music such a slick experience, everyone got excited. They created a digital music revenue stream that had not previously existed and saved the recorded music industry from the bottomless pits of Napster and the other file-sharing sites. In the process however, for whatever reasons of efficiency or plain old technology ineptitude, all the metadata that fleshes out the music got stripped away. No liner notes, no acknowledgements or special thanks and no credits. When Spotify came along, I hoped that they would do something about, but no. When Apple Music came along, I hoped they would do something about that, but no. To this day, it is almost impossible to tell from what a digital track gives you which musicians played on a track, let alone who produced it or engineered or where it was recorded or when. When we combine the richness of that kind of metadata with the surprises and delights of who sampled what, music gains a richness that is what got nerdy kids excited by it. We have to get some of that back, and encourage a return to the nerdiness of music fans. Yes, “lean back” is great for the massive passives, but I believe it’s the earnest enthusiasm of a new generation of music nerds that will really return a passion for music to the market.

 

 

Blockchain for music?

I attended an interesting event, sponsored by the Guardian and Sonos all about the future of music and whether blockchain technology could come to its aid.

My friend and co-conspirator Dave Weller of Thomson Reuters wrote a great blog on the subject: here

The Blockchain Saves the Music Industry

Spotify Buys Beats’ Analytics Provider Seed Scientific