AI+UX: A Few Notes from the Subject

Sensible AI notes.

Word: We used a number of AI fashions, every with customized standardized directions and assistants. Prompts have been tuned, versioned, and reviewed.

A number of years in the past, when AI hit the UX scene, it was both overhyped or dismissed. We approached it like another instrument: helpful, not magical. These are a couple of notes on how AI helped us cut back supply time by ~60%, from an estimated 240 hours to round 105. No headcount adjustments. No high quality loss.

Right here’s what labored, and what didn’t.

Word 1: Earlier than the first assembly

Our SOP is to be taught as a lot as we will before talking with a shopper: historical past, information, merchandise, management, and method. It takes time. Now we use AI to collect and cluster sources: market alerts, competitor strikes, and consumer sentiment. In a couple of hours, we had a top-down view. Not excellent, however sharp sufficient to ask higher questions. And higher questions construct higher strategic relationships.

Takeaway: Use AI not to impress, however to put together.

Word 2: The contract cross

We’ve learn our contract templates dozens of instances. Perhaps too many. Once we ran one by way of a legal-tuned AI mannequin, it flagged imprecise language, edge-case clauses, and areas a shopper may problem. Small points, however the sort that ship contracts again for revisions. This time, the shopper signed inside a day.

Takeaway: Let AI reveal what routine makes invisible.

Word 3: Analysis synthesis

Pages of notes are gradual to course of. We fed AI our screenshots, transcripts, and notes. It surfaced contradictions, themes, and even accessibility flags. It didn’t determine. It simply helped us see extra clearly.

Takeaway: Use AI to spot alerts. Your job is to verify them.

Word 4: Visible exploration

The temporary included customized illustrations. We requested AI to sketch early iconography. It gave us a stable begin. Hours turned minutes. Last polish was achieved manually. We don’t waste effort profitable battles that don’t want to be fought. We let AI put together vector sketches, so we will focus on making them production-ready.

Takeaway: If the first 70% comes quick, you’ve gotten extra power to care about the final 30%.

Word 5: Code, with a caveat

Mid-project, the shopper proposed an thought that will elevate the bar, but in addition enhance dev time. Scope creep. First, we wanted to take a look at feasibility. AI brokers helped draft the logic. Seven hours later, we had a working POC. However deployment stalled as a result of safety wasn’t aligned. Was it wasted work? Perhaps. However it uncovered friction we will now design for.

Takeaway: Pace exposes the place alignment is lacking. That’s nonetheless progress.

Word 6: The invisible work

Emails, assembly summaries, consumer tales. They are all vital, all gradual. AI drafted most of it. We nonetheless edited, however we by no means began from zero. It freed us to keep centered on the precise design.

Takeaway: The worth isn’t in automation. It’s in reclaiming your focus.

The imperfections

AI stumbled. Visuals wanted cleanup. The code wanted debugging. Prompts advanced day by day. However AI was by no means the ultimate voice. We learnt how to talk with AI to make it useful, one small step at a time.

Takeaway: An assistant with out ego is higher than a peer with none of your context.

Some financial savings numbers

Complete discount: ~135 hours (out of 240). Solely current platforms, used with intention.

Last word

“The model we offered right now acquired compliments from a number of groups. Nice reactions throughout.”

That was sufficient.

One final takeaway

Use AI to get you considering higher than yesterday.

The article initially appeared on UX Design Lab.

Featured picture courtesy: Mykhaylo Kopyt.