Unlocking the Potential of Generative AI: Build or Buy?
The decision of whether to build or buy generative AI solutions is not just a matter of immediate costs or convenience; it’s a strategic choice that can have long-term implications for an organization’s competitiveness, innovation capacity, and overall success.
Navigating this decision requires careful consideration of a multitude of factors.
To shed light on the most critical element in this evaluation process, we asked a panel of experienced tech and business leaders to share their insights.
We focused on one key question: What is the single most important factor to consider in the build-or-buy decision for generative AI, and why is it so vital?
Their answers, explored in this post, offer a roadmap for making choices that will position organizations for success in the age of generative AI.
Read on!
James Allsopp
A critical factor in deciding whether to build or buy generative AI solutions is organizational expertise and resources.
Developing an AI solution that is tailored for specific functionalities internally is not trivial as it needs a team that has well rounded experience in machine learning, other data sciences, and applicable AI ethics, and also has good infrastructure to train and deploy the models.
In one example, an AI focused SaaS company we were consulting had an existing AI team and proprietary data, so they chose to build their solution internally. In contrast, many startups do not have the required bandwidth and are more successful pursuing ‘buy and plug’ strategies with existing solutions.
Such engagements influence timeframes, budget estimates, and end performance. Use of in-house expertise makes the deployment of models efficient, while lacking the requisite knowledge makes it inefficient, costly, and time consuming.
Most importantly, internal resource assessments must be done thoroughly before any action is taken, as it promotes long term returns.
C.L. Mike Schmidt
One of the most important factors in deciding whether to build or buy a generative AI solution is the expertise available within your organization.
Developing an AI solution from scratch demands specialized knowledge in data science, machine learning, and software engineering, which may not be readily available in-house.
For example, in my legal practice, we considered whether to use off-the-shelf litigation software or build a custom tool.
Without the necessary technical expertise, building a reliable solution would have been costly and inefficient, making the purchase of a well-tested product the smarter choice.
Doug Gotay
When you’re deciding whether to build or buy GenAI solutions, the first step is to take a good look at your organization’s internal capabilities and resources.
Ask yourself: Do we have a person or team with the technical know-how to pull this off? Do we have the infrastructure in place? Do we have the time and budget required? And do we have a use case that justifies investing in building a custom solution?
For some use cases, such as content creation, there are a number of off-the-shelf GenAI solutions that will do the job, and very well. But these pre-built solutions may not have the flexibility and customization options you are in need of.
Your decision needs to include weighing the tradeoffs between what works best for your needs right now versus what will set you up for success in the long run.
Doug Gotay
Co-founder, Cubic Squared and an AI evangelist
Anmol Agarwal
Generative AI tends to be very computationally expensive and takes a lot of computing resources. Therefore, when deciding to build or buy Generative AI, one key consideration is to determine the cost of deploying a Generative AI solution. It’s cheaper to buy Generative AI versus building it from scratch.
In addition, building tools in-house also requires access to key personnel knowledgeable about Generative AI which could be data scientists or software developers. Organizations should ask themselves whether they are willing to invest in these key personnel.
However, when you buy Generative Ai solutions, there is also additional security risk. You need to determine whether the vendor will process your data securely.
One concern with AI is ensuring data security, so when you buy a solution, you must ensure that that solution employs good security controls and is trusted. When building a tool in-house, you have more control over your data and ensuring that the data is secure, but it can also be more expensive to do so
Anmol Agarwal
Senior Security Researcher, Nokia
Alari Aho
Budget constraints are often the deciding factor when choosing between build and buy.
Building AI from scratch requires upfront investments in talent, tools, and infrastructure. Buying a solution often comes with predictable costs, but customization may be limited.
This financial aspect directly impacts not just short-term decisions but also long-term operational efficiency. Organizations must weigh these trade-offs to ensure financial sustainability alongside innovation.
Balázs Keszthelyi
One important factor an organization must consider is the alignment of generative AI solutions with its strategic goals.
When evaluating whether to build or buy, it’s crucial to assess how well the solution integrates with the company’s long-term vision and operational framework.
For instance, if a company aims to enhance customer engagement through personalized marketing, a bespoke solution may be necessary to tailor algorithms specifically to its customer data and preferences.
On the other hand, if the objective is to quickly implement AI capabilities without extensive resource allocation, purchasing a ready-made solution could be more efficient.
This factor influences the decision-making process as it directly impacts not only the effectiveness of the solution but also the overall return on investment. A misalignment can lead to wasted resources and missed opportunities, which is something organizations should strive to avoid.
Balázs Keszthelyi
Founder & CEO, TechnoLynx
Shomron Jacob
Data security!
Public LLMs are powerful, but they also can expose a business’ data and create privacy vulnerabilities when their employees submit sensitive information through these genAI systems.
So, organizations need to be really careful whether a solution they are buying relies on a public LLM (and how that aligns with their broader data security strategy), or whether building out a private LLM infrastructure would better protect their data (while delivering more on-target results).
Getting this right is important.
Private LLM solutions that companies can build out will keep sensitive information under organizational control while still enabling their custom AI applications to do what they need it to do.
I think particularly as many organizations start scaling their AI initiatives, that ability to maintain data privacy—while generating proprietary insights–will be a lasting competitive advantage.
Shomron Jacob
Head of Machine Learning and Platform, Iterate.ai
Dr. Gregory P. Gasic
One of the most important factors is data security and compliance. In the healthcare technology industry, generative AI solutions handle sensitive patient and provider information, making adherence to regulations like HIPAA and GDPR non-negotiable.
Building a solution in-house allows organizations to design robust security measures tailored to their specific data handling protocols, ensuring full control over compliance and minimizing risk.
Conversely, buying an off-the-shelf solution requires rigorous evaluation of the vendor’s security standards and ongoing monitoring to ensure compliance with industry regulations.
A failure to prioritize data security and compliance can lead to severe legal, financial, and reputational consequences, undermining trust among doctors, remote professionals, and patients. Thus, this factor is critical to maintaining the integrity and viability of the platform.
Dr. Gregory P. Gasic
Co-Founder, VMeDx
Hiren Shah
One crucial factor when deciding whether to build or buy generative AI solutions is organizational expertise in AI development.
Developing generative AI in-house requires specialized talent, substantial resources, and time. If a company lacks experienced data scientists and engineers, attempting to build in-house can lead to delays, increased costs, and subpar results.
Conversely, buying a pre-built solution allows organizations to quickly leverage proven AI technologies, often supported by updates and maintenance from the provider. However, this comes at the cost of customization and control over proprietary data handling.
This decision must hinge on whether the organization’s core competencies align with AI development.
For instance, at Anstrex, we focus on leveraging tools that complement our expertise in competitive intelligence while outsourcing non-core functionalities to maintain agility and focus.
Hiren Shah
Founder, Anstrex
Cache Merrill
One of the strategic decisions any organization needs to make relates to the AI ownership and control of intellectual property IP.
Insourcing eliminates the issue of licensing third party solutions and, therefore, allows maximum tailoring of solutions. On the other hand, the scalability of purchased solutions is out of the client’s control, including how updates and data will be used after deployment.
In one case, a client of ours chose to design and develop an AI analytical tool that fits the exact scope of their business.
Such an approach not only saved them from incurring licensing costs but also enabled their team to make changes to the solution to meet emerging needs, saving them over $500,000 vendor costs in three years.
This decision hinges on long-term strategy—control versus convenience often defines the right path.
Cache Merrill
Founder, Zibtek
On behalf of the Techronicler community of readers, we thank these leaders and experts for taking the time to share valuable insights that stem from years of experience and in-depth expertise in their respective niches.
If you wish to showcase your experience and expertise, participate in industry-leading discussions, and add visibility and impact to your personal brand and business, get in touch with the Techronicler team to feature in our fast-growing publication.
