Hey - it’s Fiona

Enjoying the sunny weather here and putting the finishing touches on a few design projects (always satisfying). A couple of new ones are kicking off this week and next, so I’m looking forward to getting stuck into those.

I’ve noticed more founders asking me, “Can I just use AI to design the UX?” It’s a fair question — there are a lot of tools making big promises right now. But in marketplaces especially, AI has real limitations.

That’s why for this weeks topic I’ve chosen: What AI can’t (yet) do in marketplace UX — and what that means for your product decisions.

What can’t AI (yet) do in marketplace UX?

There’s a growing temptation to outsource marketplace design to AI. Tools like Uizard, Galileo, and even ChatGPT plugins can generate UIs in seconds. So why pay for a UX designer?

As much as I embrace AI in many of my processes, it’s not quite there yet when it comes to UX design. I wholeheartedly agree that AI is getting impressively good at layout and visuals, but marketplace UX isn’t just about pretty screens. It’s about understanding messy human behaviour in a complex, multi-sided system — something AI can’t yet replicate in a meaningful way.

Here are 4 critical things AI can’t (yet) do in marketplace UX, and how you can work around those limits:

1. Prioritise features based on messy, real-world tradeoffs

AI will suggest everything (e.g. filters, reviews, ratings, dashboards) because it doesn’t understand resource constraints, technical limitations, or the emotional context of a founder bootstrapping their first version.

Instead: Start with your riskiest assumptions and design around what needs to be tested, not what looks “complete.”

2. Understand trust dynamics between buyers and sellers

Trust is the invisible currency of every marketplace. Whether it’s a cleaner entering your home or a startup booking a consultant, the UX must build confidence at every step. AI can’t detect the emotional nuance of when to show credentials, social proof, or badges.

Instead: Map out your user’s fears and design friction where it helps (e.g. ID verification), and speed where it’s safe (e.g. rebooking flow).

3. Design for supply-side complexity

Sellers, service providers, vendors all have different workflows. Some want to approve each job. Others want auto-booking. AI can create a “generic” dashboard, but it can’t assess which actions are most important to each role.

Instead: Interview or observe one seller type in your niche. Then design a seller dashboard that prioritises their key goals, not just what the template suggests.

4. Balance business model and UX trade-offs

Should buyers be able to message sellers before paying? Should you allow negotiation? These aren’t design questions — they’re strategy questions with huge UX implications. AI doesn’t understand your business model, margins, or go-to-market plan.

Instead: Align your UX choices with how you’ll make money and build trust.

Bottom line: AI can speed up wireframing and surface ideas — but the heavy lifting of designing trust, prioritisation, and value exchange still falls to you (or your UX partner 👋).

—> ✉️ Reply with your questions and I’ll answer them in a future issue.

Magician‑style plugins are cool, but think strategic

Fabrice Grinda’s breakdown of “AI in Marketplaces” highlights real, founder‑level AI use cases like:

• Autocomplete & listing creation

• Smart recommendations

• Fraud detection & moderation

Tip for marketplace builders

Use AI plugins to speed up repetitive tasks (e.g. autofill listings), but reserve core flows—like negotiation, reputation, and onboarding—for strategic design that reflects your unique brand and trust model.

This week’s must‑reads for founders navigating AI in marketplaces:

Fiona Burns

Find me on LinkedIn or Book a 1:1 call

If you enjoyed this newsletter, why not forward it to a friend.

Did someone forward you this email? You can subscribe to Marketplace Minute here!