AI
Workflow Automation
Mobile App
Reducing Delays in Inspection-to-Estimate Workflow with AI
Designed an AI-powered inspection assistant to guide technicians, capture complete deficiency data, and auto-generate first-draft estimates. Reduced turnaround from 1–2 months to just days, improving approval rates and revenue conversion.
Process Highlights
Project Explanation
My Team
2 Product Designers
1 Product Manager
3 Engineers
Responsibilities
UX Research
Journey Mapping
Wire-framing
Prototyping
Usability Testing
Design Systems
Handoff
Disciplines
User Experience Design
User Interface Design
Product Design
Interaction Design
Motion Design
Tools
Figma
FigJam
After Effects
Background
Inspections often left gaps—technicians skipped fields, and estimators spent weeks chasing details before creating estimates. This slowed approvals and risked losing clients.
“Why can’t the app guide us to capture everything—and draft the estimate instantly?”
That insight led to FieldAssist, an AI-powered agent that ensures complete data capture and generates first-draft estimates within days instead of months.
Double Diamond
Methodology
Research
Discover
Problem Statement
Manual inspection workflows are slow, inconsistent, and prone to errors. Technicians often miss key data, causing delays in generating estimates. For businesses, this means lost revenue, missed upsell opportunities, and a poor customer experience—at a time when speed and precision are critical to outpacing competitors and securing jobs.
Research
A multi-modal research approach was adopted, starting with stakeholder interviews across departments to uncover key needs, challenges, and expectations. This was followed by in-depth qualitative interviews with field technicians, owner and estimator to gain insights into on-ground behaviors, pain points, and mental models during inspections.
Insights in from stakeholder and users
To understand the full inspection-to-estimate process, I interviewed:
Technicians – Shared on-field challenges such as signal loss, skipped fields under time pressure, and reliance on pen-and-paper for quick notes.
Estimators – Highlighted recurring issues with missing inspection data, dependency on technician follow-ups, and delays in creating accurate estimates.
Owners – Offered both operational and strategic perspectives, focusing on customer retention, revenue impact, and the need for transparency in workflows.
Internal Stakeholders – CS, Sales, and AM leaders stressed the importance of aligning tools with business goals and ensuring smooth client experiences.
Total participants: 24, selected to ensure balanced insights before moving forward.
Estimate: Understanding the Estimation Workflow
While analysing the estimate creation process, we identified key areas where delays occurred, with significant impact on customer retention. Here's what we found:
Estimators frequently spent 1-2 weeks preparing estimates, leading to potential customer loss if not delivered on time.
The first draft of estimates is crucial for retaining customers, as delays risk losing them to competitors.
Labor charge estimation is more complex than material estimation, as it requires an understanding of repair times and other variables.
The process also includes additional steps like obtaining vendor quotes, creating purchase orders (POs), and estimating service hours, adding to the turnaround time.
Immediate estimate drafts could significantly reduce the friction in customer experience and speed up decision-making.
These insights highlight critical inefficiencies in the estimation workflow, directly impacting customer retention and revenue growth.
Data Collection: Understanding the Technician Workflow
While examining the data collection process during inspections, we noticed inefficiencies that caused delays and gaps in information. These were the main challenges:
Technicians often missed key details during the inspection, requiring follow-up calls to complete the data.
Encouraging technicians to gather all necessary data during the inspection itself could reduce delays and improve efficiency.
Some offices have implemented Google Forms for deficiency data collection, which technicians fill out separately from their inspection reports, creating an extra step.
The deficiency data collected included various parameters like Reason, Images, Recommendation, Severity, Brand, Model, Quantity, and Size, but was often incomplete.
By incentivizing technicians, possibly through gamification, we could encourage them to fill in all required data.
Automating notifications for deficiencies would ensure that estimators are promptly informed, reducing delays in the estimation process and improving response time.
Platform Data Insights - From ZenFire
To identify inefficiencies in the inspection-to-estimate pipeline, we analyzed usage patterns and workflow data from FieldOps CRM, a product already used by our technician and sales teams.
Time-to-Completion (TTC) Analysis: Using Mixpanel’s completion metric, we analysed user flow from data collection to estimate approval, identifying delays in data collection, missing information, and the estimate approval process. Metabase was used to segment data by roles and inspection types for deeper insights.
Approval Rate vs. Delay Correlation: We tracked estimate approval rates relative to completion time. Our goal was to determine whether longer timeframes negatively impact customer approvals. Insights showed that delays beyond 3 days led to a measurable drop in approvals.
Contextual User Behaviour (Flow Funnels & Screen Recordings): We reviewed screen recordings to observe user behaviour in real time. This helped us uncover contextual pain points—such as hesitation during form inputs, missed buttons, or repetitive navigation patterns—that analytics alone couldn’t explain.
Competitor Analysis
We wanted to understand what alternatives exist in the market today for our customers, and how those experiences shape their expectations. This helped us set realistic benchmarks.
A consistent framework of 5 UX parameters was used to analyze and benchmark the user experience of 6 competing platforms.
Workflow Efficiency
Customisation flexibility in inspection forms and data collection.
Number of steps and speed required to complete a full inspection-to-estimate workflow.
Availability of auto-estimate generation features.
Offline capability and sync reliability.
User experience
Simplicity and clarity of UI for field technicians and estimators.
Mobile responsiveness and usability under poor network conditions.
Ease of navigation for non-tech-savvy users.
In-line guidance like tooltips, contextual help, and error handling.
Feature Set
Multimedia support (image/video uploads during inspection).
Voice-to-text capabilities.
In-app messaging or tagging between field and back-office teams.
Template-driven estimate generation workflows.
Integration with CRM, billing, or scheduling tools.
Analytics & Reporting
Real-time performance dashboards.
Conversion funnel tracking (inspection → estimate → approval).
Delay detection with reason codes.
Approval & Communication
Time-to-approve benchmarks.
SMS/email notifications to clients.
E-signature and digital approval support.
Customisation & Scalability
Role-based access and permissions.
Configurable forms based on service categories or inspection types.
Support for multiple languages.
Pricing & Market Placement
Subscription tiers, pricing models and flexibility in plans (monthly/annual, freemium, enterprise).
Target customer segment (SMBs, enterprise, individual contractors).
Designing for Tomorrow – AI & Emerging Tech in Field Inspections
Envisioning how emerging technologies — especially AI — can radically transform field inspections and estimations by enhancing user efficiency, reducing friction, and unlocking business value.
Refined Problem Statement
Technicians struggle to complete inspections accurately and efficiently in real-world conditions due to inconsistent network access, time pressure, and complex forms. This leads to incomplete data capture, delayed estimates, and friction between field and office teams. As a result, businesses face slower turnaround times, lost upsell potential, and diminished customer trust — making it harder to scale operations and stay competitive.
Research
Define
After synthesising all qualitative data from interviews, contextual inquiry, stakeholder conversations, and competitor analysis, we transitioned into the Define phase. This stage focused on making sense of the research by organising it into actionable insights.
Post-Up
Captured directly from user interviews, contextual inquiries, or internal team reflections. These notes highlight user behaviours, struggles, goals, and environment-specific constraints.
Affinity Diagramming
We grouped similar observations to identify common themes, user pain points, and opportunity areas. These clusters laid the groundwork for defining user needs and experience principles.
User Personas
The user personas were built to capture the diversity of roles: technicians, estimators, and owners highlighting their day-to-day journeys and pain points. This ensures the product remains grounded in actual user expectations.
Jouney Maps
Step-by-step flows capturing how each role interacts with the inspection-to-estimate process.
Empathy Mapping
Capturing what users say, think, feel, and do to uncover real frustrations and motivations.
Prioritisation Grid
Captured directly from user interviews, contextual inquiries, or internal team reflections. These notes highlight user behaviours, struggles, goals, and environment-specific constraints.
To-Be Jouney Maps
Step-by-step flows capturing how each role interacts with the inspection-to-estimate process.
Observations
Technicians often skip optional or unclear fields during inspections, creating gaps in data capture.
Estimators spend weeks chasing missing information, delaying approvals and risking lost revenue.
Current workflows rely on multiple disjointed tools (PDFs, emails, spreadsheets), leading to inefficiency and human error.
Owners struggle with fragmented dashboards and need to check multiple tabs to track job progress, costs, and risks.
Communication loops between technicians, estimators, and owners are slow and repetitive, increasing frustration.
The lack of offline-ready workflows forces technicians back to pen and paper in no-signal zones, slowing digital adoption.
Estimate rejection rates are high due to incomplete or inaccurate data, lowering client trust and approval speed.
There is no proactive system to flag revenue leaks or provide predictive insights, limiting owners’ ability to make data-driven decisions.
Estimators face pressure to submit faster but lack AI assistance to pre-draft or optimise estimates.
Owners want more strategic visibility (risk alerts, SLAs, margin impacts) but often get lost in raw data instead of actionable summaries.
Research
Develop
After synthesising all qualitative data from interviews, contextual inquiry, stakeholder conversations, and competitor analysis, we transitioned into the Define phase. This stage focused on making sense of the research by organising it into actionable insights.
User Stories
Step-by-step flows capturing how each role interacts with the inspection-to-estimate process.
MoSCoW
We applied the MoSCoW prioritisation framework to separate features into what must be built first and what could be phased in later. This gave us a clear release roadmap while ensuring development stayed focused on high-impact features.
“The process helped us cut out features that would have created friction in development and potentially delayed timelines, allowing us to deliver value faster without extending sprints.”
User Flow
Step-by-step flows capturing how each role interacts with the inspection-to-estimate process.
Information Architecture
Step-by-step flows capturing how each role interacts with the inspection-to-estimate process.
Defining AI Design Principles
I wanted to define guiding principles that balance human workflows with AI assistance ensuring trust, speed, and clarity while keeping users in control.
Human-in-the-Loop
AI assists but never fully replaces human judgment. Users (technicians, estimators, owners) always remain in control with clear override options.
Effortless Sync
Data flows seamlessly between technician inputs, estimator workflows, and owner dashboards AI ensures no duplication or gaps.
Context-Aware Guidance
AI prompts adapt based on role and task (on-site inspection, estimate drafting, revenue tracking), ensuring relevance instead of noise.
Speed Without Sacrifice
AI reduces manual entry and delays but never compromises data accuracy. Inspections, estimates, and approvals move faster while staying reliable.
Reliability & Safety
AI outputs are grounded in validated job data, regulations, and past performance never assumptions or hallucinations.
Invisible Assistance
AI stays in the background auto-filling forms, suggesting drafts, surfacing insights without interrupting human workflows.
Ideation
After creating low-fidelity wireframes, I tested them with potential users through interviews to validate assumptions in real-time before moving into full-scale product design.
“During ideation and validation, I interviewed 25+ users to refine flows and finalize the low-fidelity wireframes.”
Data Collection Made Easier for Technicians
Technicians often struggled with skipped fields, unclear inputs, and repeated re-entry of data at the end of the day. This led to incomplete inspection reports and unnecessary back-and-forth with estimators, slowing the overall workflow.
To address this, I designed an AI-assisted data capture flow that ensures technicians can collect complete, accurate data on-site with minimal effort:
Voice, Image & Photo to Form AI: Technicians can record notes or capture photos, which are automatically converted into structured fields — reducing missed details.
Step-by-Step Guidance: Inspections progress naturally through low-friction prompts, with AI surfacing relevant citations and recommendations at the right time.
Smart Prompts & Mandatory Fields: Critical fields are flagged upfront, ensuring technicians complete essential details without confusion.
On-the-Go Flexibility: Additional notes, images, or voice inputs can be added seamlessly without disrupting the workflow.
Zero Redundancy: Once submitted, inspections flow directly into estimates — eliminating rework and unnecessary follow-ups.
Working with Estimate
Technicians often struggled with skipped fields, unclear inputs, and repeated re-entry of data at the end of the day. This caused incomplete reports and unnecessary back-and-forth with estimators.
As part of ideating the estimate flow, I explored how AI could simplify and structure the process so technicians and estimators could create accurate estimates with less effort:
AI-Generated Drafts: Estimates are automatically created after inspections, reducing manual entry and saving time.
Clear Line-Item Hierarchy: Costs are broken down into categories with quantity, rate, and totals for easy review.
Deficiency-to-Estimate Mapping: Each line item is linked to identified deficiencies, improving traceability.
Flexible Input Options: Line items can be added or updated via text, voice, or image references in the field.
Streamlined Approvals: Transparent status indicators (In Progress, Approved, Declined) simplify tracking across multiple estimates.
Contextual Summaries: Each estimate includes a clear service summary, reducing client confusion and back-and-forth.
Empowering Owners with Real-Time Insights
Owners need a clear view of their business performance, from revenue growth to team efficiency, without digging through reports or spreadsheets. This dashboard was designed to centralize key insights, helping owners make informed, data-driven decisions in real time.
As part of ideating the owner dashboard, I explored how AI could simplify tracking and forecasting, giving owners the tools to oversee operations effortlessly:
Revenue Overview: Track won revenue, open opportunities, and pending follow-ups at a glance.
Performance Analytics: Monitor efficiency, job completion rates, and revenue generated per technician.
AI-Powered Forecasting: Predict upcoming revenue based on deficiencies detected during inspections.
Opportunity Management: Identify where estimates are approved, rejected, or pending, improving clarity and accountability.
Centralised Data: A single source of truth for inspections, estimates, and team performance to reduce miscommunication.
Handovers
Deliver
Once the ideas were validated, I focused on making the designs production-ready, ensuring scalability, responsiveness, and smooth collaboration with development teams.
“I ensured the final designs were production-ready, scalable, and consistent across mobile, tablet, and desktop platforms.”
Design System & Guidelines
Built a reusable component library (buttons, forms, layouts, etc.).
Established design tokens for color, typography, and spacing.
Created a single source of truth for consistency and scalability.
Responsive Design
Designed layouts for Mobile, Tablet, and iPad breakpoints.
Adaptive designs tailored to both field technicians and managers.
Tested edge cases for accessibility and usability.
Developer Handoff & QA
Prepared detailed documentation and interactive prototypes.
Collaborated closely with developers during implementation.
Supported QA with flow testing and resolved design inconsistencies.
Handovers
Impact
Qualitative Impact
The app consolidated fragmented workflows (paper, spreadsheets, PDFs) into a single AI-assisted mobile environment, improving coherence and reducing friction.
Technicians and estimators now have guided workflows, reducing guesswork and ensuring critical data is captured.
First-draft estimates are auto-generated, enabling stakeholders to focus on refinement rather than starting from scratch.
The solution enhances transparency and trust - everyone (field, office, owners) sees the same data and status.
It elevates decision-making by surfacing predictive insights (e.g. rejection risk, conversion bottlenecks) rather than just raw numbers.
Quantitative / Business Impact
Turnaround from inspection: estimate dropped from 1-2 months to just days.
Potential improvements in approval rates and revenue conversion, as delays are minimised and estimates are higher quality.
Faster decisions and reduced follow-up cycles mean cost savings in manpower and operations.
Better data completeness and accuracy reduce the volume of estimation rejections or revisions.
More streamlined processes unlock capacity - more inspections/estimates can be handled in the same time.
ZenTrades CRM
Field Scheduler