Right now Artificial Intelligence and Machine Learning are hot topics. Who is better suited to do work in that area? Large companies or startups? Large companies, because they have huge data set or startups because they can innovate faster even though they lack the data that large companies do? For example, Google has huge data set in GMail, so can a startup compete in that space with limited data set? One of the recent a16z podcasts deals with this topic and during the conversation an interesting point regarding innovation in large companies was mentioned. That was something I experienced in a previous startup and is worth expanding.

It has nothing to do with technology. It is about how a large company can setup a culture of innovation. When a startup is opening to users, it has to work to make the experience as great as possible. Else, you would have some wonderful machine learning algorithm sitting in the cloud and no one using it. You could either have a great architecture or be in business. A small company has freedom on using the right security policies, on defining how they use data, on how they store data and on how they sanitize it.

Imagine a large company starting a small group for innovation. This could be a like a lab for incubating ideas or like a Horizon 3 group. While creating such a group, if you blanket them with the same Terms and Conditions as the larger company, then it would be restricting the freedom of innovation. The team essentially would spend time jumping over operational hoops (We won’t let you go to production without the company blessed security review) and technical hoops (You have to use one of the approved languages and databases for development) instead of focusing on getting a product out.

Steve Blank writes

Horizon 3 is where companies put their crazy entrepreneurs. (Inside of companies these are the mavericks you want to fire for not getting with program. In a startup they’d be the founding CEO.) These innovators want to create new and potentially disruptive business models. Here the company is essentially incubating a startup. They operate with speed and urgency to find a repeatable and scalable business model. Horizon 3 groups need to be physically separate from operating divisions (in a corporate incubator, or their own facility.) And they need their own plans, procedures, policies, incentives and KPI’s different from those in Horizon 1

Then, you may ask: Won’t the mother ship suffer from reputation loss, if the internal startup messes up? Aren’t these checks and balances required? There is a better way which I was part of at a previous company. We had a small team of innovators which built a startup, served customers and actually moved money. The company in its wisdom gave complete freedom to the development team to innovate. When the entire company ran in a data center, we ran off AWS and no one batted an eye-lid. When we wrote the UI in Angular, no one said no. The only condition was that, it would not have the company brand till we proved success. That decision freed us from working to get check marks on a corporate checklist and instead focus on customer problems. Our energy was spent in running experiments and validating various hypothesis which is what an external startup would be doing.

When companies start innovation groups, they should also empower them and trust them to make the right decisions. Else, it would just be lip service to innovation.