Head in the clouds: Why AI matters for mid-market companies
The foundation for AI has been laid in the last few years with uptake set to accelerate owing to , potential to scale, and growing talent to support it. Jan Beitner, Assistant Director at Inflexion focused on Data & AI, talks about key long-term drivers of this dynamic area.
When you’re on the cloud, scaling is easy. A lot of data systems and machine learning applications such as personalisation often require real-time data. Your systems need to pull out pertinent data quickly, but if you are hosting on premise, then your systems are likely to be overwhelmed. This is particularly the case if you are growing quickly, ever increasing the usage of your existing infrastructure. Slow system responses can lead to a subpar user experience or even worse to outages, which in turn means missing out on potential revenues. Effortless scaling is crucial to many AI systems which is one of the reasons so many businesses (including 90% of the Inflexion portfolio) have started to migrate to the cloud.
Cloud hosting is also easy; you don’t need to maintain your own systems anymore. A small company needn’t build a huge IT stack; you can now buy “managed services” off the shelf and have specialist software vendors maintain them. This allows you to focus on your core business with a better and cheaper IT stack. It’s easy to get onto the cloud, with system integration often taking mere days or even hours, far less than the months often required for on-premise solutions.
The growing popularity of managed services is leading to commoditisation not only of software and infrastructure but also of the most advanced AI around language understanding and image recognition, enabling even small companies to embrace this game-changing innovation. The start-up Hugging Face runs an online platform where data scientists can shop cutting-edge AI and datasets. A few years ago, the extortionate cost to create those AI models made it the preserve of global tech giants; now it’s accessible to most. It has truly disrupted the market.
That said, while the vast majority of companies are starting to use the cloud, most aren’t making the most of all it has to offer. For example, they often still retain core functionality like legacy data warehouses on premise. There is a lot of headroom for efficiency gains even for those companies that have taken the next step and decided to deeply integrate with cloud hyperscalers such as AWS, Azure and GCP.
The latest research in AI is maturing from ‘let’s play’ to having real business impact with commercial relevance. The early days of advanced, deep-learning-based AI saw it applied to games like Chess and Go, and more recently it’s played a pivotal role in semi-autonomous vehicles. Now we see advances in drug discovery being supported by AI.
We can now also see AI creating completely new images just based on short text descriptions, even winning art competitions. AI like Dall-E 2 from OpenAI has the potential to disrupt marketing by generating personalised creatives at only $0.02 per image. GitHub Copilot is another example for a new type of AI that already generates cash flow for their owners. It writes code for developers and added more than 400k subscribers in its first month paying $100 per year. They have the potential to fundamentally change existing business models and the routines of many knowledge workers.
In order to reliably deliver high ROI, AI solutions must be assessed thoroughly and used responsibly. Many have heard about the ethical risks AI can entail, with text generating models for example posing potential for misinformation, spam, phishing, and more, according to cynics. Those are clearly business risks to consider, but responsible AI isn’t restricted to that.
At the end of last year, real estate marketplace Zillow announced it was closing its ‘Zillow Offers’ unit after incurring a $569 million loss. It had used its proprietary data to create an algorithm but without correctly assessing and testing it. This incredible loss illustrates the dangers of running with unvetted AI – and the need for a real intelligence (i.e., human) overlay. In fact, responsible AI practices such as making machine learning model decisions transparent almost always boost returns and is therefore becoming popular.
The supply of AI Talent is increasing – though remains well below demand. Just five years ago, you had to hire a computer scientist or mathematician and then train them on data and AI for your company’s needs. Universities are now adapting, with business schools in particular offering courses focused on business analytics or big data. According to the Institute of Advanced Analytics, more than 100 degree programs alone in business analytics exist in the US in 2022 compared to less than a dozen ten years ago. This means there’s more relevant talent, particularly from recent graduates, with precisely the skills businesses seek. These people can readily analyse data with cloud-based tools. This creates a powerful combination of in-house talent and managed services. It’s a huge shift, and while there remains a shortage of talent owing to exceptional demand, the supply is at last increasing too.
You can no longer hide from AI; it is becoming a disruptor in myriad areas. Firms that have laid the data, talent and infrastructure foundations for engaging with AI will reap the rewards as uptake accelerates.
All Inflexion portfolio companies, regardless of size or ownership stake, have full access to our value acceleration resources covering digital enhancement (including data and AI, technology and cybersecurity), international expansion, add-on acquisitions, ESG, commercial strategy and talent management.