A pilot project quickly shows whether a new technology is relevant for your company. But danger lurks around an unexpected corner: too successful a project has more disadvantages than advantages.

Is AI right for your business? Or are you considering managing all your data in a different way? Maybe it’s time to provide more automation via low-code and no-code? Whatever the plan, no one likes to jump in at the deep end blindly. A promising idea is followed by a first small project: das conceptual proof (PoC).

PoC

Such a PoC shows how realistic the technological implementation of a new concept is. A company, whether together with partners or not, is looking for a clearly defined project in which the new technology can prove itself. In this phase, the focus is not on radical changes in business processes.

For example, an organization could replace part of its internal technical support with a chatbot. This allows the company to explore the possibilities without immediately releasing the experiment to the public. Some accountants may also have access to some automation tools to simplify recurring aspects of billing. For data, a PoC can consist of transferring some data to the cloud and correlating it with other data there.

Results in two weeks

If you do the latter, you might meet Joris Van den Borre. He is the founder of Tropos.io and works, among other things, as a partner of Snowflake on implementations at companies. Snowflake’s data cloud concept is popular and also appeals to us. Anyone who works with Snowflake opens up data in the cloud, where it is available for people and applications according to the zero-copy principle, not only internally but also externally if desired.

“A success story can be built in just two weeks,” says Van den Borre. “Perhaps the biggest problem is that your problem becomes too successful.” That sounds like riding, but it’s anything but. Moreover, it’s not unique to Snowflake.

entry-friendly

There are two major pitfalls with PoCs of new technologies: the fact that the new solution will be cloud-based in most cases, and sophisticated management. First, let’s take a look at the cloud aspect.

The cloud enables companies to set up a PoC within weeks or, who knows, even days. The result can also be very useful right away. Finally, the low entry limit and high scalability is one of the biggest advantages. Of course, you pay to use it, and the limited scope is typical of a PoC. In principle, no immense budget is provided to immediately serve the entire company with the pilot project.

You ask, the cloud is running

Sometimes things can go wrong. Perhaps the findings from the pilot project are so relevant that other departments want to get started right away. Or maybe it’s immediately apparent that a chatbot can pull up a lot of relevant information from across the organization. If you allow usage to escalate at this PoC stage because everyone wants to get their hands on the exciting new features as soon as possible, you’ll be hit with the bill (literally) later.

Physical restrictions apply to your own server, but these do not exist in the cloud.

Joris Van den Borre, Founder and CEO of Tropos.io

“Snowflake can deliver what you ask for, as long as you ask for it,” says Van den Borre. “Owning your own server has physical limitations, but these don’t exist in the cloud. That requires discipline. If you use capacity in a disciplined manner, you get a lot for little money.” The downside is clear: those who loosen the brakes easily consume beyond their means. A PoC thus becomes a money hole.

culture change

Of course, scaling is the ultimate goal of any project, but before that can happen, the corporate culture must also change. The consumption model requires you to think carefully about who gets access to what, and for what reason. Anything is possible in the cloud, but not everything is immediately desirable.

Too fast and too many problems can also cause problems with locally running projects. For example, think of low-code and no-code automations. They are useful, but they also need to be managed. Ideally, admins will be willing to run a variety of apps securely, but that’s not a given. When everyone suddenly starts creating their own automations, it becomes a mess that’s hard to navigate. This in turn brings with it security and compatibility issues.

Better slow and good

Don’t rush things, set goals and stick to them. Yes, the new project will be ready within two weeks and can serve the whole company within three weeks, but it doesn’t end well. The necessary cultural adjustments are necessary, discipline is necessary and the planned use must be matched by a budget. Only then does the scaling take place.