Build or Buy AI: Could you find one to buy?
One of the recent trends in consumer preferences has been the transition from cooking at home to ready meals. People all across the world are looking for higher-quality cuisine without having to make it from scratch, such as me. I was fond of Waitrose’s (British supermarket chain) ready meals during my stay in London. Aside from an unwillingness to spend time in the kitchen (honestly, that was my sole objective), one cause for the transition has been customers’ desire to try new culinary combinations and cuisines.
Ready meal versus homemade is simply a projection of the famous make-or-buy decision that corporate executives have faced during their MBA training. In the corporate world, executives are confronted with this dilemma where it’s often a matter of cost, time, quality, and performance. For example, you could either choose to build your artificial intelligence (AI) system or use one that already exists on the market.
As AI continues to evolve and grow in complexity, more and more people are beginning to ask the question — could you buy an AI instead of making it yourself? After all, investing in a ready-made solution can save you a lot of time, money and effort while ensuring you’re getting a high-quality product at the same time. There are numerous companies out there that are offering pre-built AI solutions specifically geared towards solving specific problems within different industry sectors, so it really couldn’t be easier to find a solution that meets your needs.
But is this the right approach for your business? And what are the benefits of creating your own AI system as opposed to buying one from a supplier? In this article, we’ll take a closer look at this topic and provide you with the insights you need to make an informed choice about whether to buy your AI or build it yourself.
Before you make any decisions one way or the other, it’s a good idea to get some clarity around exactly what your business goals are and what problem you’re trying to solve. Once you’ve identified the problem, you can determine whether an existing off-the-shelf solution is the best solution or whether building your own AI will be more cost-effective and give you more control over the outcome. Let’s take a look at some of the pros and cons associated with each option so you can make an informed decision about the best approach for your business.
There are two main benefits to buying an off-the-shelf AI solution rather than trying to build one yourself.
First of all, these systems are built by highly skilled and experienced developers, which means that they’re far more advanced than a basic off-the-shelf solution in terms of functionality, reliability, performance and scalability. In many cases, these solutions are developed by companies that have myriads of experience in the AI space, so you can be sure that you’re getting a product that has been thoroughly tested and that is capable of meeting your business requirements. This also means that you won’t need to spend time and money hiring and training a team of developers to develop your system from scratch — you can simply buy the solution that has already been developed and save yourself a lot of hassle and expense in the process.
The second major benefit to buying an AI solution is the cost — most systems cost several hundred thousand dollars to purchase outright, so they are much more affordable compared to the cost of developing your software. While it’s true that these solutions are fairly expensive, it’s still considerably cheaper than the cost of hiring a team of developers and hiring them to write a bespoke piece of software for your business.
The main downside to buying an off-the-shelf AI solution is that you’re limited by the features and functionality of the solution — you won’t be able to add any additional functionality to the system to customize it to meet your needs, and this will often result in you having to sacrifice certain features to gain access to other features. For example, you may decide that you need a system that can read a specific document and automatically generate a response. Still, you won’t be able to do this with an off-the-shelf solution because it’s designed to read a different type of document instead. As a result, you may find that the off-the-shelf solution isn’t suitable or powerful enough for your needs, in which case you’ll have no choice but to develop your custom solution.
Overall, though, the advantages of buying an off-the-shelf AI solution far outweigh the disadvantages, so it’s usually the best option for businesses that don’t have an existing team in-house that can create a custom software solution for them.
The big question is: Could you always find a ready-to-go solution? The answer depends on the industry! It would be strange indeed if digital agencies could find no off-the-shelf AI software for analysis, lead generation or customer relations management. Meanwhile, it is more difficult to find an out-of-the-box solution, for instance, for an advanced technology company occupying a specific niche (cancer research, autonomous robotics etc.). That resembles the fact that you can’t probably find traditional dishes (such as kaburga dolması, i.e., stuffed ribs 😋) as a ready meal in supermarkets — certainly not everything people could cook at home would be sold with packaging and instructions in a supermarket (even though Waitrose’s ready meals section used to provide me with a generous selection). For more complicated recipes it’s better to order the food or cook it yourself and then enjoy the result — the more you want complex, the less you find ready-made offers in the supermarket.
Finding a suitable AI product for manufacturing companies must be harder than finding your favourite traditional dish in the prepared form since each manufacturing plant has its own set of processes, machinery and workflows. Some of those processes are so custom and complex that it practically doesn’t make much sense to search for a product that has already been designed for general purpose and therefore may not match your specific requirements exactly. In such cases, a customized approach is usually the only viable option for a business, and so the search for a ready-made AI solution comes to an end. This is why you see more custom-made solutions for industrial manufacturing rather than purchased products.
Let alone finding an AI product, it is challenging even to scale the internally built solution. A good example is Microsoft’s Project Bonsai implementation at PepsiCo for Cheetos production. The project was a huge success but the scaling process must have been a struggle because of the complexity of creating simulation frameworks, a prerequisite for Reinforcement Learning (Project Bonsai’s main technology), using operators’ expertise.
The same applies to most other kinds of use cases too. Therefore AI adoption at manufacturers is still in its early stages (according to Gartner, only 21% of companies have active AI initiatives in production).
This situation results in a downsized market and thus smaller margins for the AI ventures in the space that are trying to solve the problems. CBInsight’s recent report “Factory analytics and artificial intelligence for advanced manufacturers” features only 9 companies whereas their Natural Language Processing report demonstrates about 70 different companies. It’s logical to assume that AI companies pursue markets other than manufacturing because they need less customization. As a result, manufacturing companies are left with a smaller selection of ready-made AI solutions to choose from and are forced to develop their custom software.
Whether you choose to develop your own customized AI software or rely on a third-party solution is your choice to make. But if you go for the latter option could you find one to buy?
