Skip to content
Kelli Meyer
BOISE ··· ··:··
Contact

QTEX AI

CUTTING DECISION OVERWHELM FOR HOBBYIST ENGINEERS

UX Designer & Researcher2025Client redesign sprint
The design sprint, on paper — 5-second test, Crazy 8s, and low-fi wireframes. Tap a card to bring it forward.

Context

QTEX AI is an electronics-component platform — "enabling engineers to build better products, faster." I ran a focused design sprint to redesign the search-and-selection experience inside their PartWise tool, and delivered it as a leadership-facing business case with a live, high-fidelity interactive prototype, presented to QTEX AI's leadership over Zoom.

Problem

PartWise asks people to choose from more than 30,000 components. For a hobbyist that's paralysis: too many options, technical jargon that locks out non-engineers, hours lost toggling between vendor sites, and nagging compatibility worries that end in abandoned carts. I anchored the work in a persona — Alejandro, who loves building but has no formal training — and a sharp problem statement: choice overload and jargon stop hobbyist builders from finishing a purchase.

The reframe

The issue wasn't a shortage of options; it was the absence of a path through them. If I could lower the cognitive load at the decision point — and build trust into the recommendation itself — the drop-off would take care of itself.

Approach — three core features

I built the redesign around three moves:

  • Smart Search — natural-language input ("I need a red LED for my Arduino breadboard project") with AI suggestions, so you don't need the right part numbers or jargon just to start.
  • AI-Powered Curation (the "Super Vote") — a best-match recommendation carrying a confidence score and a plain-language "Why this?" rationale (compatibility plus how often the part is used in similar projects), to cut analysis paralysis.
  • Persistent Cart — an always-visible cart that tracks selections and keeps purchasing under the user's control.
Before / after of the part-selection flow — the original plain “Find electronics” search on the left, the redesigned natural-language chat interface (with annotated message bubbles and clear context) on the right.
Before / after: the original keyword search vs. the redesigned Smart Search chat — surfacing results inline and making the AI’s role legible.

Research — three iterations

I tested a deliberately rough low-fidelity prototype with a small moderated group — my instructor, two design peers, and my husband, a working engineer standing in for the professional user. The feedback contradicted my assumptions in exactly the way good testing should, and drove three iterations into the high-fidelity prototype:

  • Iteration 1 — Make the context clear. Testers couldn't tell where the results lived (a popup? a page? the chat?), confused the project name with the part name, and didn't realize they were talking to an AI. Fix: surface results as inline chat responses with message bubbles, and split the project-name header cleanly from the part name. Afterward, all four testers immediately understood the chat context that had confused them before.
  • Iteration 2 — Build trust with data. The clearest signal: "Can't only trust AI data — I need the factory datasheet too." Professionals verify against authoritative specs. Fix: a collapsible Technical Specifications section (collapsed by default) with a Download Datasheet (PDF) link — progressive disclosure, so a beginner gets simple guidance and a pro can verify everything without leaving the flow.
  • Iteration 3 — Respect user control. The original "Add to Project" language created friction — it implied saving data and commitment, which raised security concerns, and the path to actually buying was unclear. Fix: rename it "Add to Cart," rename the BOM widget "Shopping Cart" with "items saved temporarily" microcopy, and make "Save Cart as Project" explicitly optional. "Cart" reads as temporary and user-controlled, and it makes checkout the obvious next step.
“User Testing Insights: What Users Told Us” — three findings from moderated testing: confusion about the AI’s involvement, professionals wanting factory datasheets, and concerns about saving project data.
What testing surfaced — confusion about the AI’s role, professionals needing authoritative datasheets, and worries about saved data — the three findings that drove the iterations.
“Iteration 2 — Technical Transparency: Building Trust Through Data” — pairing user distrust and verification needs against the fix: a collapsible technical-specifications section with datasheet downloads and progressive disclosure.
Iteration 2 in detail — layered technical transparency: collapsible specs and a datasheet download, so beginners get simple guidance and pros can verify everything without leaving the flow.

Outcome

I packaged the redesign as a leadership-facing business case: a projected 10–15% conversion lift (framed conservatively), with recommended pilot KPIs — conversion rate, time-to-purchase, cart abandonment — and the trust-and-retention upside: fewer returns from pre-purchase compatibility checks, lower support costs, and higher customer lifetime value. I presented the interactive prototype and the business case live to QTEX AI's leadership over Zoom, with a three-month validation-and-rollout roadmap (starting with a 15–20 user test across hobbyist and professional segments) as the recommended next step.

Postscript: since the sprint, QTEX has taken its business in a somewhat different direction — but its current site now implements the core chat-interface features this redesign recommended.

A slide from the leadership business case — “Business Value: Conversion” — laying out the funnel impact, reduced decision friction, a projected 10–15% conversion increase, the reason drivers, and recommended pilot KPIs.
From the business-case deck: the projected 10–15% conversion lift and the recommended pilot KPIs — conversion rate, time-to-purchase, cart abandonment.
“Competitive Advantage: QTEX AI Differentiation” — a table contrasting incumbents (Digi-Key, Mouser, others) with the redesigned QTEX experience across catalog, guidance, and technical depth.
Competitive advantage: how the redesign differentiates QTEX from incumbents like Digi-Key and Mouser — hobbyist-friendly curation and confidence-building, without losing the technical depth pros expect.
“Next Steps — Recommended Path Forward” — a three-month phased roadmap: validate with users, technical-feasibility assessment, phased beta rollout, and continuous iteration.
The recommended path forward — a phased three-month roadmap: validate with 15–20 users, a technical-feasibility review, a phased beta rollout, then continuous iteration on real usage data.

Reflection

This sprint reset how I work. I went in assuming a clean, simplified recommendation would be enough; testing proved me wrong while it was still cheap to change. My honest takeaway: Lean UX is how you move, Rich UX is how you think — move fast with messy low-fi prototypes, but dig into the why (security, industry norms, how people actually read a word like "cart") before jumping to a fix. Don't fall in love with the first idea.

The live high-fidelity prototype — chat-based Smart Search, the AI “best match” with its “Why this?” rationale, expandable specs, and the persistent cart. Try it.