Organising for AI-by-design
Part one of the Prosus AI series on how to achieve AI-by-design product and business innovation
What are the most successful AI use cases you know in your business, product or field? And why do you define them as successful? They've likely made a significant impact on a business metric by taking a process or customer proposition and improving them. Take radically optimised search, much cheaper logistics or highly relevant question answering, for example. While these applications and their impact are impressive and important, they often do not fundamentally change or future-proof a business, product or company.
This may sound counter-intuitive, as AI has become synonymous with innovation, but in reality, the vast majority of implemented use cases are for incremental process optimisation. It makes sense to allocate a large share of your AI team to work on incremental applications, because risks are lower as both impact and feasibility can be approximated — but not your whole team.
Spending 100% of the data science resources in your organisation on solving problems of existing customers, products and processes, means that you run the risk of missing AI applications that can put you at the forefront of disruption and stay ahead of your competitors.
At Prosus, we launched the Squared Initiative and developed an approach to work on AI innovation in parallel to business-as-usual incremental AI. The goal of the Squared Initiative is to find ways to accelerate the path towards radical, yet practical applications of AI that solve real customer problems. It places us in the middle ground between solving current problems users are facing (incremental, user pull) and taking an R&D innovation approach (too long term, tech push). The journey toward this middle ground is not a linear one, and we have many lessons to share — which is why we are launching this blog series on AI innovation.
In our Prosus AI Squared Initiative, we accelerate the path towards radical, yet practical applications of AI within our businesses
In this blog series, we're starting a conversation on what AI-driven innovation really means — and what you need to be successful, based on the lessons from building the Squared Initiative. Even though an increasing number of how-to guides are written to help businesses uncover applications of AI, we found that most of them focus on what we at Prosus Group call incremental applications. Google, Spotify and others have written blogs and created guides to get teams started with AI and optimise workflow to reduce the average time to production. Our focus is on AI-by-design innovation and we will detail and share our five-step approach in our next blogs:
- Explore user needs and AI trends to understand big opportunities for innovation;
- Define which 'mission' is most important to solve first — where user needs could meet AI potential (we use the term 'mission' to define an initiative and set it apart from a regular project; it is the unit of attention for the Squared team);
- Concept how to radically solve this mission, building on the unique strengths of AI and an in-depth understanding of your customers or business;
- Test to quickly de-risk concepts — both technically and in terms of desirability;
- Deliver impact at scale.
It takes a carefully organised, dedicated team to translate a high-level goal as ‘AI-by-design innovation’ into clear and testable missions. There are many ways to do so, but here we highlight the main choices we made when launching our Initiative and synthesised them into four takeaways.
Takeaway 1: Innovation requires a different organisational set-up
Create an autonomous, multi-disciplinary team that can prototype non-incremental ideas to solve real customer problems
The objectives of AI-by-design innovation are more ambitious, while the chances of succeeding are lower — and this requires a completely new way of working and organising. How to organise your innovation team may not be the first question you expect when discussing AI innovation — especially since most of the field is focusing on research breakthroughs, new algorithms and innovative tools.
But translating the latest technological innovation into business innovation is about much more than how fancy your algorithms are. This also makes for an important distinction between the Squared Initiative and an R&D team: AI innovation does not originate from scientific innovation alone. It’s not the quality of Tiktok’s recommendation algorithms alone that earn the social media app the title of ‘AI-as-the-product’; it’s the perfect alignment between those algorithms, the short video format and the understanding of the target audience. The orchestration of all these different aspects ask for a multi-disciplinary AI innovation team.
Because the challenges that you face when working on AI innovation are so different from those that we know for incremental applications, you want to have a dedicated, stand-alone team that has the autonomy to solve them. This is where the Squared Initiative is positioned. This might seem obvious but is easily overlooked as most mature organisations are designed for stability and predictability at scale, which makes it harder to prototype non-incremental concepts in rapid iterations.
Takeaway 2: Innovative thinking can be trained
Build skills and capabilities where needed: train team members to understand and leverage each other’s expertise
When launching an initiative like Squared, you may be worried that you don't have the right talent in your organisation to staff the team. When launching Squared at Prosus AI, an early sentiment was that we had to get a team together "that would also be hired by Google to design their new products." This seems to suggest that there is a single, specific profile of people out there that is able to work on innovation; you simply need to hire them for the job.
But a year into our Initiative, our conclusion is very different: innovative thinking is a skill that can be trained in many people across your organisation. Just because data scientists or user researchers work mostly on incremental topics on a daily basis doesn't mean they cannot conceptualise and de-risk innovative business ideas. By creating the right conditions for innovation to thrive, incremental practitioners can turn into innovative tinkerers. These conditions include:
- The freedom to take risks: allocate team members to the Squared Initiative on a full-time basis to free them up from operational tasks and the corresponding short-term goals (e.g. OKRs);
- The environment to learn from others: mix different experts with the curiosity to learn from other people's skills while building on their own expertise;
- The balance between new and comfortable: take team members out of their geographical or functional field of expertise to allow them to see their own organisation and its customer proposition with a fresh pair of eyes.
Not every data scientist will be comfortable working on risky AI concepts that may never make it to production; not every service designer enjoys feeling like a rookie in the field of AI. Yet plenty of them do and can be trained to become core contributors to your Squared Initiative.
Our team combines people from multiple geographies, as our businesses operate in India, South Africa, Russia, and many other countries. We unite colleagues working on similar topics in different sub-businesses in our centralised Initiative, to maximise the diversity in thought and chances that solutions will be successful across multiple markets. They rotate in and out of the team for a full-time allocation of three or six months. The composition of the team has varied over time, depending on the needs of the current mission. The core of the team is formed by at least two data scientists and two service designers or user researchers, led by a ‘mission control lead’.
Innovative thinking is a skill that can be trained
Takeaway 3: Leadership needs to take a portfolio view on its AI use cases, each with a different risk level
Ensure that leadership at different levels in the organisation is bought in to the Squared Initiative's philosophy and the inherent risks in working on future-looking concepts
Saying that you're starting an AI innovation Initiative may sound sexier than it is, given the true commitment it takes from leaders at different levels in the organisation. Working on future-looking ideas always requires determination and patience. It's not only the daily team that needs to get trained to become disciplined innovative thinkers: supporting leadership also needs to train themselves to adopt a mindset and common language around the topic of AI innovation. Overall, we find it useful to think about concepts that come out of the Squared Initiative as part of a portfolio of use cases. The largest share of a company's use case portfolio is lower risk, more incremental; the Squared Initiative adds the riskier use cases with a potential for 10 times the impact.
In this portfolio-framework, senior leadership becomes the investor that looks at each of the Squared Initiative's concepts and asks: would I spend another two weeks of resources to advance this concept? Is it non-incremental yet feasible enough? Is it bold enough? And in two weeks, what additional certainty do I need in order to continue to fund this concept? Essential here is a shared understanding of what innovation truly means in their context, and what an acceptable timeline is between sketching a high-level concept and proving impact at scale. We'll talk about this process of proving impact early on in a later blog.
Managers lower in the organisation experience the shift in the AI portfolio, to include non-incremental use cases, more directly. They take a risk when freeing up scarce data scientists and researchers on a full-time basis for an initiative that won’t impact this quarter’s business metrics (and when impact does come, it may not even be directly in their line of businesses). Setting up a successful collaboration with product and data science leads involves more than setting the right incentives: it's all about inspiring them. Even when these managers spend their time making sure that teams produce their share of the planned number of LEGO blocks for this quarter, most of them will keep a close watch on what's happening in the field. They want to feel proud and inspired to know that there's a successful team working on designing and testing the future of their company.
The Squared Initiative adds riskier use cases to the broader portfolio, with a potential for 10 times the impact
Takeaway 4: Listen, ask, and listen again
Involve everyone in the user research, to ensure that you are working back from the customer and their needs at all times
In an innovation Initiative, you want to be able to get feedback about early-stage concepts. Will our users love our solution? Will we be able to build it and integrate it into my existing products? And will the business case work out? You therefore need to optimise your team for breaking down a concept into its main hypotheses, quickly testing the riskiest assumptions and learning from early feedback. We’ll talk more about user-centric AI innovation in the next blog in this series.
One way of embedding quick learning into the design of your team is by involving the whole team in the user research. It is tempting to outsource the conversations with users to the designers in the team, who are the experts here. But it is the conversations with customers that show the true potential and limitations of a concept from the Squared Initiative, and this is essential information for designers and data scientists alike.
In part one of this Prosus AI Innovation series, we've stressed the importance of AI innovation as a mindset that's achievable for many organisations. We've highlighted the four main takeaways from building our Squared Initiative, namely: 1) innovation requires a different organisational set-up, 2) innovative thinking is a skill that can be trained, 3) leadership needs to take a portfolio view on its AI use cases allowing for higher-risk use cases, and 4) everyone in the team needs to be involved in the user research and testing to get quick feedback about concepts. We will continue to build out this blog series to highlight other insights from our Initiative, when we answer questions such as:
- How to combine AI and service or product design? How to let these two disciplines meet and ensure that AI is a radical enabler for innovation rather than a solution looking for a problem?
- What is the fastest way to go from wild idea to validated product? How to quickly de-risk AI-based concepts, without actually investing the time to build them?
- How to inspire organisations to design and build for tomorrow's users, using tomorrow's data and technology?
Stay tuned for the next edition and feel free to get in touch with us at email@example.com to continue the conversation on what AI innovation really means, and how to achieve it.