AI Walkthrough
Reducing Risk and Doubling Productivity for Roofing Field Technicians
In 2025, we launched AI Walkthrough at Zuper to make it effortless and safer for roofing technicians to capture job documentation while on-site.
I led research, design, and product definition, working closely with engineering and explored new frontiers in practical AI application in Field Service.
AI Walkthrough was the first step in transforming field documentation within Zuper, rethinking how technicians capture and share job context on-site.
This required us to create a framework to build a truly intelligent experience that works seamlessly and becomes an assistant, both for field and back office.
What difference did we make?
Quality of notes improved by
40%
from inconsistent, half-filled notes
to clean, professional summaries.
Field Techs only spend
10 minutes
as opposed to 30 minutes
prior to launch
We eliminated the friction of manual note-taking on rooftops by letting technicians talk naturally while recording photos or videos. AI automatically transcribes, summarizes, and even translates their input and creates a clean, professional note (with minimal edits needed).
Preview
What prompted us to reimagine field documentation?
Manual writing = Unsafe + Inefficient
Long checklists, quality assurance notes, and photo attachments were mandatory but time-consuming.
Technicians reported high cognitive load: balancing ladders, tools, and phones.
Risk of incomplete documentation led to frustration in back office teams.
Back-office team reported that some notes end up being too ambiguous or incomplete, making the notes have confusing context at times.
Typical Journey to Roofing Projects
Roofing Job Categories
Inspection, Roof Repairs, Emergency Services, Solar Installations,
Documentation Touchpoints
Inspection, Material Procurement, Repair work, Quality assurance, Job completion
But, all these documents lacked …
Connection
Notes and photos often didn’t line up, leaving field service managers guessing what the documentation referred to
Continuity
When multiple technicians worked on the same job, their notes rarely linked together.
Accessibility
Field techs wrote notes in their own language or shorthand, which back office teams couldn’t always interpret.
What can we do?
We knew we couldn’t ask technicians to type more or fill longer checklists while standing on a roof. The solution had to reduce risk, save time, and still deliver complete documentation for managers.
So we focused on three things
Shift from typing to talking
Capture observations on the fly
Automate quality of output
Summarization and translation by AI
Scalability & Future-Proofing
Create foundation for AI in FSM
Experimentation
We explored multiple directions to validate both usability and field adoption in both camera, and checklist - where field technicians have to document the most.
Early findings
After internal experiments, we rolled out an early beta to a select group of customers. We paired this with Mixpanel event tracking to see how Walkthrough was actually being used in the field
Video over photos: Technicians expected Walkthrough to capture video, and not just photos. A video would be much easier for them to record on site instead of having to manually click each picture. (few would still prefer having the ability to click pictures during a video, so we've added that as well).
Dual-camera mode: Managers requested a way to see both the technician and the work being done for accountability and proof-of-presence.
Launch
Handsfree field documentation
Technicians simply talk while capturing photos or videos, and AI turns their narration into clean, professional notes
AI-Assisted Checklists
Instead of typing long answers, technicians can speak their responses directly into checklists. AI transcribes and summarizes, saving time while ensuring completeness.
100% Adoption
All Field technicians started using AI Walkthrough mode to document all of their work. This meant hundreds of notes that were standardised, summarised, and translated with Zuper AI.
What difference did we make?
Quality of notes improved by
40%
using AI walkthrough
Field Techs only spend
10 minutes
for documentation
We eliminated the friction of manual note-taking on rooftops by letting technicians talk naturally while recording photos or videos. AI automatically transcribes, summarizes, and even translates their input and created a clean, professional note (with minimal edits needed).
Heads up! This feature is now live :D
Thank you for your curiosity
Feel free to reach out to me to explore opportunities, seek design advice, or just say hello
Copy my email