Should I build or buy artificial intelligence (AI)?

• Intro

• Questions to ask

• Pros and cons of building AI solutions

• Pros and cons of buying AI solutions

• Other considerations

• Conclusion

The question on everyone’s mind is no longer whether or not they should adopt AI technology, but rather, how should they implement it? You yourself may be asking “Should I build or buy AI in my current place of business?” In this article, we’ll explore several considerations that could help you decide whether you should consider building an AI system in-house or consider buying an AI product from a vendor.

To get or not to get. That is no longer the question. Before we get to discussing factors in building or buying AI for your organization, there are a few questions you, your leadership, and stakeholders should be asking yourselves. Let’s quickly go through some of them.

Let’s not get caught up on obtaining this technology all because all the cool kids are doing it.

It is very likely AI can be of great value and assist with challenges your company may be facing but before we decide to build or buy, it may be worth asking to figure out which route to go. You should ask specifics to the “why” such as what problems are you trying to solve, at what capacity can AI assist in addressing and solving that problem, and maybe, just maybe, can I solve the problem without AI? We can provide you with plenty of reasons why you should use AI, but at the end of the day, you should evaluate how AI would most benefit you and your company and which approach would be the most fitting.

Sure, we all have smart people we work with and they may very well be able to build your AI system, but even then, is it worth it? You’ll find that these questions aren’t necessarily in any order of importance. Most of these actually go hand-in-hand. Depending on the why, can also help you figure out whether it’s worth investing the time in building or if buying an off-the-shelf product can satisfy the need (usually at a lesser cost).

This is often the make it or break it. What is the time to implement? Do you have enough resources to meet that deadline? Sometimes building in-house can take months or even years to build. Is that something you’re willing to wait for?

If the previous question wasn’t a deal breaker, this usually is. According to

, the cost of AI solutions can cost between 0-$40,000 per year for third party AI software or $6,000 to $300,000 per custom solution.

Well that makes it easy, let’s just build! Before you decide to proceed, think about how much it costs for the talent and expertise, compute resources, maintenance, and not to mention the opportunity cost with the amount of time it will take to make. This can easily add up to millions of dollars.

Now that you’ve done some self reflecting and asked yourself some tough questions, we can go into the pros and cons of building or buying AI.

• Customization:

  Assuming you have the resources to build in-house, you can take advantage of building exact solutions that you need rather than packaged solutions that may be filled with unnecessary features. This can result in cost savings in the long run.

• Flexibility

  :

  Along with customization, in-house builds can provide a high level of flexibility allowing for modifications and ad-hoc requests. This can also ensure proper integration with current systems and the freedom to build to your company’s liking. This will also allow you to avoid any unnecessary “lock-ins” to contracts.

• Intellectual Property & Patents:

  When you build, opportunities for Intellectual Property (IP) ownership and patents may occur, which could provide competitive advantage.

• Sensitive Information:

  If you’re working with extremely sensitive or proprietary data, such as health or financial industries, it may make sense to build your own AI due to the added layers of liability and regulations.

• Lack of expertise:

  If you don’t have the expertise or the able talent within your organization to build, this could stop you in your tracks. Before you consider employing existing talent to learn and build software, you should carefully evaluate the level of effort required and understand if it would be a wise investment as it could end up becoming very costly both in time and money.

• Maintenance:

  Speaking of costly, if you decide to build in-house, that would mean you will be responsible for the system maintenance and updates to ensure you stay relevant and on the bleeding edge of technology.

• Time commitment:

  Even if you happen to have the expertise, building AI is quite time consuming. Though there are resources available such as open-source software, the amount of time it takes to build can exceed the desired results. Challenges in training leading up to quality production can take years to overcome. Years in which many cannot afford.

• Lack of training data:

  Even with both time and expertise, some vendors will have a library of training data that can place you at a significant disadvantage. Open source tools can help to level the fields, but with more complexity requires more data that is not always easy to come by. You can always buy data, but at a cost.

The following section will go over the pros and cons of buying AI as opposed to building it in-house.

• Compatibility:

  You can task the vendors with the dreaded task of compatibility and integration with other systems. It’s likely that they’ll want your business and will make it happen.

• Cost savings:

  Vendors have the advantage of spreading their R&D and operational costs over their growing list of clients as opposed to you incurring all of it trying to build in-house.

• Time savings

  :

  Another advantage of purchasing an AI solution is the time saved, particularly when the vendor already has data scientists and software engineers who have put in the work on labor-intensive tasks of training machine learning models.

• Incentives:

  As the number of AI providers grow, many still in early stages, vendors will be willing to offer additional services at a discounted price to acquire new business. In some cases, you can be grandfathered in and take full advantage of their offerings before their prices go up, increasing the value and incentives of the service.

• Quality assurance:

  Quality assurance (QA) often is built into the commercial packages and services, leading to higher quality and stability for their clients. Also, one less thing to worry about internally.

• Canned solutions:

  The number of packaged offerings and available solutions are growing but often don’t meet all the demands of a business, let alone provide unique solutions. The option for custom solutions often cost more money and require more time to build.

• Lack of compatibility:

  Before deciding to buy, make sure to check the compatibility of the AI solution with your existing ecosystem. Overlooking this aspect can lead to a lot of issues, a pause on the process, and may become a costly fix.

• Learning curves:

  It is quite possible that your vendor may use you as a guinea pig to be involved in their R&D process. Though this will lead to overall better outcomes, you may feel the growing pains from the lack of product optimization and unnecessary installations, costing more time and money.

Making a decision on whether to build or buy won’t be as simple as having your pros outweighing your cons on a chart. Make sure to do your due diligence and research and make sure to review with your stakeholders to ensure that it fits your organization’s overall strategy. It never hurts to have a conversation within your organization, as well as being transparent with your vendors, to see if they truly can provide the right solution for your team. It also doesn’t hurt to try it for free, if a vendor happens to offer a free trial or a lower cost option.

Another consideration would be to have a hybrid approach in which you can build when needed (complex or customization) and buy when you need simpler solutions (for the common, repetitive use cases).

There’s no doubt that AI can provide any business a competitive advantage through finding new opportunities, improving efficiencies, and providing innovative and real-time solutions. How to achieve those capabilities can be quite the undertaking depending on your available resources. Rarely, no one single off-the-wall solution can provide solutions to all your AI needs and challenges, but can still provide value that you may not be able to achieve yourself. In the end, it’s up to you to make the informed decision to elevate your organization to the next level through AI.

Can you think of any other questions or considerations to add? We would love to hear it in the comments! You can also contact us at

to discuss if

is a good fit for your company’s AI journey.