Are there any GenAI use cases where you would prefer in-house solutions over vendor-supplied ones or vice versa?
Sort by:
Exactly, the Vendor can provide insight if needed via training, how-to, and guidance. Anything being pulled down will likely be used for knowledgebase, summarization, or even grounding for your new or existing LLM prompts or scripts. Some of this and more could be PII, HIPPA, etc. Creating the separation and or governance of how and where inside or outside resources will be used would be ideal to set up and adhere to.<br>
I think the decision will hinge on maturity. Some areas like ITOps or DevOps will mature faster since most organizations use them. On the other hand, specialized areas like biomedical research, which require specific data, might lean towards using vendors. These vendors work with multiple companies and have a handle on data integrity. So, instead of building their own solutions, they might prefer a vendor’s. It'll really be a case-by-case decision, depending on the results and the specificity of the use case.
Adding to that, if something is unique to your organization and there's no generic solution out there, especially if it involves a mix of factors and needs to integrate with other systems, then going in-house instead of using a vendor might be the way to go.
I'd like to say all of it. It's going to infiltrate everywhere whether we want it to or not. In-house development is a skill I think most employees will need, either coding all the way through to assembling or guidance or training. So, all of the above.
I think we would prefer a vendor base from a lot of the more generic IT solutions for Gen AI. But obviously, within the business, it's much more about who's the best so it's really different.
when the data is private or the internal IP offers significant competitive advantage / differentiation.