Bear in mind that Amazon already has all the pieces in place to offer ML-driven recommendations, and I would be surprised if it didn't offer that option once this gets out of beta.

It's possible ML-driven recommendations may surpass the recommendations of StitchFix's "personal stylists" in terms of accuracy. And unless "personal stylists" is just a market-facing name for "algorithms," ML-driven recommendations by Amazon would certainly surpass StitchFix in terms of efficiency and operating costs.