The word Data spelled in ASCII over a gray technical background

Worth Reading – Enterprise AI Has a Data Problem, Not a Model Problem

The poor results likely stem from the data you’re working with.

The weakest link is not mysterious. It is the customer record with three versions of the truth, the policy binder that was never retired, the metric no one can trace, and the agent that was given access to all of it because the pilot looked great.

https://rodtrent.substack.com/p/enterprise-ai-has-a-data-problem

If I’m being honest, every time I hear someone say they don’t like Copilot because ChatGPT or Claude is just better, I have to wonder about that. The models available inside Copilot aren’t that different, but there is one big difference. Copilot has access to all your email, OneDrive documents, SharePoint files, etc.

That’s been a selling point for Microsoft, but it might also explain why the results differ. Your data might be making the results worse than when you use a different tool that isn’t connected to your Microsoft environment.

What is your AI using as its source of truth?

Similar Posts

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

To respond on your own website, enter the URL of your response which should contain a link to this post's permalink URL. Your response will then appear (possibly after moderation) on this page. Want to update or remove your response? Update or delete your post and re-enter your post's URL again. (Find out more about Webmentions.)