© 2023 v_labs innovation GmbH.Read our Imprint or Privacy Policy here.
In order to perform well in this high-paced, technology-driven business environment, corporate venture units need to take certain measures right now. If they do not, they risk investigating the wrong search fields, potentially betting on the wrong ideas and missing interesting opportunities. They will also be less efficient in the work they do, thereby wasting valuable resources. The following article provides some advice on how to avoid this trap and raises some questions for further investigat
In our last blog post about AI, we looked at how AI will influence the work of corporate venture units. This time we go a step further and hypothesize how corporate venture units should react and what concrete measures they should take. Disclaimer: The following blog post addresses some very big topics and not all of them can be discussed in full depth here. So please view this article as a starting point - we hope to continue the discussion in coming publications, workshops and personal interactions.
Our last blog article formulated three hypotheses about how AI will change corporate venturing. We argued that it will make the process more efficient, allow for better decisions and alter the ideal skill setup within a corporate venture unit. We also argued that there will still be tasks to do for people working in corporate venturing in a couple of years.
But, and this is the argument of this article, in order to perform well in this high-paced, technology-driven business environment, corporate venture units need to take certain measures right now. If they do not, they risk investigating the wrong search fields, potentially betting on the wrong ideas and missing interesting opportunities. They will also be less efficient in the work they do, thereby wasting valuable resources. The following article provides some advice on how to avoid this trap and raises some questions for further investigation.
In our whitepaper “Your Innovation Needs a Strategy” from 2019 we argued that the sheer amount of potential business opportunities provided by digitalization can lead innovation units to paralysis, in which it is hard to find a starting point - and that a well-formulated innovation strategy with clearly defined search fields for new ventures was a key ingredient for overcoming this problem. AI reinforces this argument. The sheer buzz around AI with a constant flow of news articles about technological breakthroughs and new startups makes it very hard to focus the efforts of a venturing unit. Also, the current speed of developments can make it feel as if there is no point in trying to do anything in the AI field, because tech giants and AI scaleups simply have more resources and the better starting point than a “normal” firm.
An innovation strategy with clearly define search fields serves as a north star for the activities of the venturing unit and reduces noise by labelling large parts as out of scope. Thereby it provides a clearer vision of the in-scope field and will allow us to see niches that are not visible when only assessed superficially. Ideally, it also features an inventory of assets and capabilities (i.e. unique tools, knowledge, processes, technology) of your organization which you can build upon when venturing. The resulting unfair advantages provide a key lens for investment decisions. So if your venturing unit does not have a clearly defined strategy, now seems like a good time to develop one.
Luckily, most venturing units do have a strategy. But, as many fundamental parameters have changed in the last half year, a re-evaluation of some of its parts is advisable.
First, some innovation fields that were previously out of scope due to technological limitations or because of a lack of strategic fit might have become attractive. Yet it might be the other way round: Some focus fields might have lost their appeal and/or potential. And to make it even more complex, some search fields might not change in attractiveness overall but will have to be approached from a different angle or with a different set of capabilities.
To make it more concrete while also showing the complexities involved, here are a few examples. Venturing fields that have become more accessible include areas like industrial process improvement or healthcare and diagnostics. Both fields used to rely on large amounts of data and intelligent algorithms, which rendered them unattractive for companies without proprietary datasets and without the resources to develop their own statistical models. Now, with the availability of third-party AI models, the entry barriers have been lowered considerably and companies that do not pose over data and large amounts of resources, (but for example have a good use case) could enter the field as well. On the other hand, fields that have become potentially less attractive in their original form include everything related to standardized content delivery or one-on-one advice (e.g. psychological, business, tutoring and teaching) as AI could in theory deliver the same output at a much lower unit cost. However, viewed from an AI-embedded perspective, the fields are probably more attractive than they used to be, and the disruption potential is sizable. Another point of view is provided by Chalmers et al (2021) where the Authors argue that AI will likely make all the fields more interesting for venturing that requires the search for technical solutions (i.e. by replacing multiple traditional experiments), such fields where insights can be generated from social media and sentiment analysis and fields where many assumptions need to be tested quickly.
Does that apply to all industries and company sizes or can these statements be generalized? Probably not, because the correct answer to search field attractiveness will always depend on the exact context - something we cannot provide in this blog post.
Yet, what an innovation strategy needs to do in any case, is to specify the starting point very well and then move on to how a company wishes to approach existing or new search fields from the new angles required. This means specifying the assets and capabilities with regards to AI (knowledge, processes, application fields, data from sensors across a deployed product range) that are currently available and outlining which capabilities a company intends to build up in the near future. Re-Evaluation of search fields is usually not something a venturing unit “just does” in isolation, but the result of a process. As such it is a great tool to engage in a productive discussion with your top management and other relevant departments. Not only should you discuss goals, targets, time horizons, search fields and financials but also the ethical and legal implications of AI. Data privacy is an obvious fix-starter here and a discussion on how much risk the organization is willing to take, should be pro-actively stated.
Once you have updated your search and focus fields, you should apply these learnings and decisions to your current venture portfolio: Do they still fit? Do you need to challenge some of your recent investment decisions? While the result of this process might lead to some hard-to-swallow outcomes, we believe that it must nevertheless be done. We also recommend going one step deeper. For each and every venture in your portfolio, the backlog and product roadmap should be critically evaluated with an AI lens. Some features that have been fully designed might not be worth implementation anymore because there are much smarter solutions available now. Conversely, new opportunities might have become available that make better use of your resources. As stated in the intro, these are some big topics and we are merely touching the surface. When going into more detail here, an even stricter separation of the interrelated aspects of AI (e.g. AI features of “normal” ventures vs. AI-focussed ventures) should definitely be high on the agenda
While we argue that in the long run, the ideal team composition will change (also see below), it is fundamental to build as much knowledge in your current team as possible. To be precise, every team member should experiment with the available tools and build a basic understanding of applications, use cases and limitations. Additionally, a small group of more tech-savvy employees should dive deeper into the matter, understand the underlying technology (for example how does a transformer model work? How does semantic search function?) and stay up-to-date on current developments. Who does what should be made explicit and knowledge should be shared regularly. In the end, every team member should be aware that AI is now likely going to be an omnipresent topic in some form, regardless of which opportunities a venturing unit tackles - in that way, AI will be for employees of venturing units what “digital” was 10 years ago.
After having ensured that your team proactively engages with the topic of AI, you should try to move beyond the very easy and obvious use cases and try to do more sophisticated POCs. What do we mean? Could you use the open AI API to give some brains to your existing ventures? Could you develop a Q&A system that leverages the whole information that is stored on your intranet? Could you train an open-source model for a specific use case of yours? Such POCs can only be done as part of a well-staffed project with some budget. And while these will likely not be business models, they will provide valuable insights and learnings for the real business models to come. The obvious question of how to assess whether a specific POC is worth spending money on is, unfortunately, a complicated one that would require a deep dive in a separate publication.
In our last blog post on AI, we argued that AI will lead to teams having to rely more on senior knowledge and experience. But team compositions are not easily changed. So corporate venture units should look at their strategic HR plans for themselves and for their ventures and therefore reassess their previous hiring goals from seniority, but also from a skills perspective. Finding high-class, senior people will probably require different hiring strategies and whether your HR department is up to the task is not certain. And while you might not want to hire all those people immediately, you should build awareness now that you might need a headhunter or a specialized recruiting agency to eventually fill the positions. You should also see whether the people you need fit into corporate salary bands or whether you need to look into those.
When it comes to whom to hire, Haefner et al (2020) argue that for AI-based projects to succeed, the teams working on the solutions need to generally comprise both technical employees and domain experts. This would imply that the future venturing teams will (still) consist of venture managers and business experts, UX-designers, marketers and software developers, yet that also accommodate AI experts such as data scientists, data engineers, and machine learning engineers.
This has two implications: First, these people need to be recruited and integrated into the venturing organization in an optimal way. Having more highly skilled people with very deep domain knowledge might mean that your organization will have to move from a pyramid structure with team leads managing junior members, to an alternative model such as fully self-managed teams. Whether you want to base your AI people closer to your developers or your data analytics people is another question to be addressed. As other corporate departments might also be thinking about building up AI capabilities, it is worth aligning to avoid building up similar resources at different places.
Second, the more complex team setup of venture projects with an AI component has implications for managing these. While the optimal model has not yet been devised, leaders of venturing units should pay close attention. At the very least, more effort needs to be spent on alignment, exchange and knowledge sharing. In the long run, a venturing unit will have to experiment with new approaches in project management. You guessed right, what kind of approaches make sense in this respect is another topic that requires some additional pages…
The general complexity of AI for innovation means you cannot deal with the whole topic all by yourself. Building an external network takes time and hence you should start right away. Multiple ways to go are available. You can look for R&D partnerships with universities, make selective investments in AI startups or foster strategic alliances with other companies. Just keep in mind to be clear on what to expect from your partnership. Is its financial return, better sensing capabilities (i.e. staying on top of relevant developments), reduction in (financial) risk, support in sensemaking (i.e. understanding how technology will influence you), or simply PR benefits that you are expecting to gain from those partnerships? If done right, your innovation strategy should help you with this question. Once you have clear expectations, you should draft an “offer” for potential partners. E.g. you should have a clear idea of what doctoral students or startups would need to bring to the table and what they could expect to gain in a partnership with you. In any case, we advise you to allocate enough resources (time and money!) for these partnerships. From experience, partnerships without significant resources are neither taken seriously (on both sides) nor do they tend to produce any benefits. To make the argument for such investment more convincing: a recent HBR article argued that companies who rely on more external resources and partnerships are more likely to be successful in the application and use of AI technologies (D’Silva, V. & Lawler, B. (2022).!
Finally: Once you have got your own head around what you want to do, start educating the people outside of the venture unit on opportunities, drawbacks and limitations so the whole organization is aligned on AI topics. We promise it will make all further discussions easier. Your practical use cases, low-hanging fruit demonstrations and innovation strategy will again come in handy :-)
No corporate venture unit will be left untouched by the developments in AI. While some of the implications are years out, innovation managers need to start taking actions right away. These include (re-)formulating the innovation strategy and re-assessing the current venture portfolio. They should build knowledge in the existing team and adapt hiring and staffing plans as well as project setups. Additionally, external partnerships need to be built with concrete benefits for all involved parties. While these rather general prescriptions are certainly not a panacea, we believe starting to act now is absolutely imperative for every corporate venture unit.
