My favourite web-comic is xkcd.com. And probably the most useful comic is this one:
Can AI show you the data relationships across your various applications? Yeah, I guess. Kinda. It probably can’t stop the 'test and fix whirlpool' devouring your project budget. Here’s why:
AI is good at regression and classification. Very good, even. But these techniques are not useful for situations that they haven’t seen before. Data may (or may not) be shared across applications for reasons that are buried in years, perhaps decades, of poorly documented, messy corporate decisions. This means that the training data is unlikely to exist and, therefore, wasn’t applied to the model you are using.
How about Generative AI? GenAI results against structured data are a bit rubbish. GenAI solutions need to be combined with a Knowledge Graph first. (Let’s gloss over the cost of training and running GenAI in production for now...) At DataJPS, we can help you can fill that prerequisite. Relait will scan your application data and automatically generate a Knowledge Graph of the data relationships. And it does that in a secure and privacy preserving way, dovetailed neatly into your existing enterprise RBAC systems and policies.
Why does Relait succeed where regression and classification fails? Relait tells you exactly what the quality of the relationship is. It is both accurate and deterministic. No ifs, buts, or maybes. You can fine tune the Knowledge Graph depending on your individual project needs - deterministically.
AI can’t be the substitute for your own, business specific Knowledge Graph. It’s not one or the other. Instead, use your Knowledge Graph to enhance and augment your AI projects.
We’ve described before that building a Knowledge Graph using conventional methods is hard. Given that AI won’t do it properly either, why not give us a chance to show you how quickly we can bring business impact to your data projects. Get in touch using the contact form below.
