AI Manual Hub Design for HVAC Technicians by Aleyna ÇatakAI Manual Hub Design for HVAC Technicians by Aleyna Çatak

AI Manual Hub Design for HVAC Technicians

Aleyna Çatak

Aleyna Çatak

An AI manual hub for HVAC technicians. The user is on a job site, the manual is three hundred pages, the phone wins.

Client: None. Manuel AI is an existing platform. Check their website.
Scope: UX Research (persona, problem statement) + Flows + Wireframes + Hi-Fi + AI Interaction Design
Timeline: 1 Week
My role: The whole arc, from persona to pixel

The Problem

Fokke, an HVAC technician, installs and repairs products from many brands and needs quick access to many manuals. Carrying them is tiring; searching PDFs on a phone mid-job is worse. Manuel AI is an all-in-one digital manual hub: ask a maintenance question, get the exact procedure back. The design problem underneath: AI answers take seconds, and a technician standing on a ladder experiences seconds differently than a designer at a desk.

Constraints

The core interaction is waiting — the honest version of an AI product's UX, designed instead of hidden.
One persona, taken seriously: everything on screen justified against Fokke's pain points and work context.

The Approach

Decision 1: show the pipeline, not a spinner. While the answer generates, the interface narrates named agents doing real stages: Memory Expert reviewing context, Intent Detector, Knowledge Navigator, Answer Generator.
Decision 2: design the wait as a feature with three competing options. Instead of one waiting treatment, the study explores three: animations, rotating HVAC facts ("a clogged filter is the #1 cause of system failures"), and an optional trivia mini-game with a score. Rejected the pretense that latency doesn't exist; the options make the wait respectively tolerable, useful, or fun — and testable against each other.
Decision 3: flow chart before wireframes, persona before flows. The sequence was strict: problem statement and persona, then flow charts, then wireframes, then a redesign pass, then hi-fi.

The Range Moment

The waiting UI mirrors how retrieval-augmented AI pipelines actually sequence — understand intent, search the corpus, generate — because the person designing the screens understood the architecture underneath them. The transparency isn't decorative; it's a truthful system diagram wearing a chat interface.

The Outcome

A full concept arc from persona to hi-fi, plus three testable waiting-state treatments. Everything delivered in 1 week.
Like this project

Posted Sep 1, 2026

Designed AI manual hub for HVAC technicians with testable waiting-state treatments.