Field Notes: AI has a UX problem | Lakhveer JajjField Notes are thoughts and musings from building Moselle — real-time<br>shower thoughts, building out loud, and saying the things others are already<br>thinking. Very opinionated, but grounded in what I’m seeing in the field.<br>That felt worth sharing out to the universe. No more, no less.
And a quick level set before we start: I think LLMs are one branch of AI,<br>not the whole thing. Treating them as the whole thing is a disservice to<br>other elements of AI, like linear programming, time-series forecasting, and<br>constraint programming. But most of the “AI” being marketed today is really<br>just LLMs. So when I say AI in this post, I really mean LLMs.
A few weeks ago I caught up with a founder friend over drinks. She runs a<br>consumer brand, and lately, founders building AI products for commerce<br>businesses keep reaching out to her — not to pitch her, but to ask how to<br>sell to someone like her. Somewhere into the second glass she flipped the<br>question on me: why is Moselle having success<br>approaching brands while everyone else she talks to is struggling?
I’ve been chewing on my answer ever since. The honest version is that we<br>stopped fighting user behavior and started designing around it. The longer<br>version is this post, and it starts with two things I believe about this<br>moment. The AI revolution is real: look at a benchmark like<br>Finance Agent v2 and watch frontier<br>models climb on core financial-analyst tasks. Top scores hover around 60%,<br>though under the benchmark’s stricter all-pass scoring, where every detail<br>must be right, none clears 51% (hold that thought). And the<br>models are no longer the moat. Every serious player has access to roughly the<br>same intelligence. What the technology can’t do, no matter how many<br>parameters you throw at it, is change user behavior.
That’s why the headlines are so contradictory. Some companies report massive<br>AI spend and consumption, while<br>MIT’s “GenAI Divide” report<br>found that 95% of enterprise AI pilots showed no measurable P&L impact. Both<br>can be true. Building Moselle, a planning platform for fast-growing consumer<br>brands, has given me a front-row seat to why. What I see on the ground is<br>three problems, and none of them are about the models.
Problem 1: “Trust me bro” is not an adoption strategy
Users are ingrained in doing the work themselves. Their tools, their<br>routines, their sense of professional competence — all of it is designed<br>around them producing the output. Telling someone that AI will now do that<br>work isn’t a feature announcement; it’s a massive UX shift, and we’re mostly<br>asking people to make it on faith. On top of that, the industry’s default<br>interface for all this power is an empty chat box. Type what you want the AI<br>to do. That’s the product.
An empty chat box is a black hole from a UX perspective. It’s the blank Word<br>document problem: you can write anything, which is exactly why most people<br>write nothing. A blinking cursor doesn’t tell you what the system is good at,<br>where its edges are, or what to try first. It hands the user all the<br>ambiguity and none of the guidance.
The non-deterministic nature of AI makes this worse. When the same question<br>can produce slightly different answers, users don’t experience “creative<br>flexibility.” They experience uncertainty, and uncertainty reads as<br>untrustworthy. Closing that gap takes heavy education on when to trust AI<br>and when not to, and that can’t be solved overnight. Even platforms built<br>AI-first still require users to develop judgment about when to hand work to<br>the AI and when to step back. We see this friction at Moselle all the time,<br>with customers who want to adopt.
Human nature being what it is, adoption will be gradual. But I’d argue the<br>interfaces are a big part of why the rollout feels slow, and it could<br>accelerate sharply the moment someone ships the killer app whose interface<br>isn’t purely chat. Until then, gradual is still fast by any historical<br>standard, but not fast enough from a VC perspective. Which means the name of<br>the game isn’t speed of adoption. It’s retention, and consumption economics<br>that prove the work is genuinely cheaper and better than what a human can do.<br>That’s a proof you earn over quarters, not demos.
Problem 2: The commerce ICP isn’t the early adopter
Coding tools exploded first for an obvious reason: founders and engineers are<br>the natural early adopters. They’re building the thing — of course they’re<br>going to sell it to their friends. If your addressable market is startups and<br>tech companies, the early-adopter pool is enormous.
The numbers back this up. In the<br>Anthropic Economic Index,<br>computer and mathematical work (largely software engineering) accounts for<br>about 37% of all Claude usage, despite those roles being a sliver of the<br>workforce. Meanwhile the Census Bureau’s Business Trends and Outlook Survey<br>puts AI adoption in retail trade around 14%, near the bottom of the table,<br>with professional services and finance at the top. The<br>Fed’s...