Thoughts on the near future of AI in B2B products

I spent the last week adding AI to my product, Flourish Commerce. We do UX and conversion rate optimization audits for eCommerce stores. Now, I'm impressed by AI just generally speaking. As someone who's been working with computers my whole life it's mind blowing that you can upload a screenshot of an eCommerce site and the computer can describe what it's looking at. However. When you ask that same AI how to improve the design or optimize for conversion rates, you get some pretty bland responses. Some of it's actionable, but a lot of it is incredibly basic or just straight up irrelevant. And this isn't the only domain I've found responses like this. Luckily, we've been doing our audits completely manually for the past few months and have developed a database of best practices along with supporting facts, figures, and sources with further information. Combining that human-curated database of best practices with the capabilities of AI is what creates something really special. When you feed AI with expert content, it's more than capable of using that information to assess how well the design follows the best practice. Then it comes up with relevant, actionable advice that's specific to the brand, the design, and the guidelines you've provided. With just a little direction—prompting, I suppose—the output is easily 10x better. We're going to use AI to make expert, customized UX advice accessible to every brand, and I think that's something that will apply to a ton of domains in the near future—at least until GPT-6 or 7 just embeds ALL the knowledge and makes hand-curated databases irrelevant, lol. Anyone else building AI workflows that are augmented by expert content? How much does it improve outputs vs. base models?