Media Query Source: Part 48 - Techronicler (India digital magazine); Important factors to consider when making build-or-buy Gen AI decisions

• Techronicler (India digital magazine)
• Important factors to consider when making build-or-buy Gen AI decisions
• More than one build-or-buy viewpoint should be considered
• Data security and maintainability are top of mind for me

My responses ended up being included in an article at Techronicler (January 27, 2025). Extent of verbatim quote highlighted in orange, paraphrased quote highlighted in gray. Above image from cited article.

The query responses I provided to Techronicler on January 14, 2025:

Techronicler: What is one important factor an organization must consider when deciding whether to build or buy generative AI solutions? Why must this factor influence the decision-making process?

Gfesser: This consideration can be taken from two different viewpoints: an organization determining whether they should build or buy a generative AI product to support their business, or an organization determining whether to buy another organization that offers a generative AI product.

The reason I'm calling out these two viewpoints is based on my experience building software and data products myself or with my teams, as well as considering purchases of organizations offering products built atop generative AI models.

In both scenarios, one concern is data security. AI product offerings run the gamut from open source and self-hosted to proprietary and third-party hosted, with the former providing more control than the latter, providing potentially lessened risk. I've discovered software and data product security leaks in the past, but this latter extreme makes discoveries of this type much more challenging due to lack of visibility.

Another main concern of mine is maintainability. One advantage of buying a generative AI product to support a business is the maintenance package presumably bundled with it. However, as a business buyer I've personally considered purchasing several organizations that offer generative AI products, only to discover that many of these are built as a very thin layer atop the APIs of a single AI model, significantly increasing dependency risks.