What to think about - Build vs. Buy for the Automotive Industry

It may feel like time to rehash the old question: When should I build, and when should I buy? In the age of AI, this question becomes even more complicated.

Part of this discussion was prompted by a conversation I recently had with an automotive company that has tasked their internal IT team with creating an AI system that had been in process for more than three years,and still wasn’t functional. Internal technical teams often have an enormous queue and even when a project does make it to the top of the queue, the team often must drop everything to deal with an internal emergency before getting back on track. I asked the executive: “If you had to do this all over again, knowing what you know now, what would you do?” You can probably imagine the answer I received.

I would like to help others faced with this predicament, especially those in the Automotive Industry: When does it make sense to build internal software, and when does it make sense to buy internal software? And what are all the considerations?

1.      Costs – This is a big one, and costs are also nuanced. There are initial costs, maintenance costs, scaling costs, hiring costs, tool costs, and/or storage costs. Which of these costs you’ll incur when experimenting with AI software for your own organization will depend on several factors.

2.      Time – This is another big one. How quickly do you need software to be ready to go? Do you need it in a week, a month, a year,or a decade? It’s easy to underestimate how much time it can take to build internal tools by your IT or Tech Team. Sometimes tech teams measure time to finish in terms of development – but may forget to factor in time to test, deploy, and iterate based on feedback and results.

3.      Customization – Figuring out how much of the application will need to be customized is another key decision point. Some customizations are easy for a software vendor to do but others may take more work. It’s also important to consider the fact that the more customizations are implemented,the farther away from a the tried-and-true, tested solution you are going.

4.      Expertise – Building an AI platform or tool from scratch takes a significant amount of expertise and experience. Of course, some organizations have staff with the levels of needed expertise ready to go.However, it can be difficult to hire and retain such staff. We at Humaxa, for example, are fortunate enough to have PhD-level AI experts that have been building AI applications for years – because that’s all we do. Other organizations are not as fortunate.

5.      Scalability – When focused on building software tools internally, it will be critical that whoever built the initial software tool shares their processes, methods, and documentation – in case they don’t stay with the company indefinitely. It’s also possible that the original builder will move to a different role and leave a gap. A vendor that builds AI software platforms likely has redundancy and backups in place because their business depends on it.

6.      Integrations – The ability for a platform to “talk to” other internal systems may be very important, but the build vs. buy decision could go either way here. If a platform is built internally and there are enough talented resources whose job it is to build integrations, the decision to build internally may make sense. If an external vendor is contracted to add an integration to their existing platform, they may have already built similar integrations– or not. A vendor may have a team dedicated to integration support – or not. It’s always important to ask.

7.      Security – When building custom software, it can be built from the ground up to adhere to internal and external security standards. However, a vendor’s off-the-shelf  or customized-for-you software may also follow security standards. The important thing is to ask the right questions: How often are penetration and vulnerability tests run? Who are they run by? Are they run by a reputable third party? What were the results? Were any critical or high incidents detected? If so, what has been done to mitigate those risks?

8.      Support and Maintenance – When something goes wrong, it’s essential to have people who can step in and start troubleshooting quickly. This may be an internal, on-call team or a vendor-supplied support team.Either way, it’s important to solidify the support process and know who is ultimately responsible for maintaining the platform.


Like all build/buy decisions, automotive industry companies must weigh the pros and cons of each option and determine which option makes the most sense. Costs and time are relatively easy to measure and may be a good place to start. Like a close colleague of mine always says: “There are no good or bad decisions – only math.”

What do you think? Have you had positive or negative experiences with either building or buying yourself? What other considerations do you think organizations should consider?


Carolyn Peer

CEO & Co-founder, Humaxa


Recent stories

Thinking of Experimenting with AI? Document your current processes.

When experimenting with AI, how one might want to go about documenting current processes

“I’ve been tasked with figuring out how my organization can and should experiment with AI. Great! Now what?”

How to choose a pilot project to experiment with AI in your organization

"I’m thinking about experimenting with AI. What do I need to know about involving stakeholders?"

What to consider when starting an AI pilot project in the Automotive Industry, specifically around stakeholder involvement