Designing and shipping a full operations platform for a construction company
I joined Veles as a graphic and brand designer. When the team's existing tools failed to support their workflow, I was asked to design a replacement. I researched their day-to-day operations, designed the full product in Figma, and shipped it as a working web and mobile application without writing code by hand
Role
Graphic Designer → Product Designer
Timeline
Apr 2025 - Ongoing
Scope
Web and mobile, full design and build
Context
The team had tried general project management tools before, and none of them held
Veles is a home renovation company of about 20 people, split between office staff and subcontractors in the field. Tools like ClickUp had been adopted and abandoned within weeks. Most coordination happened through WhatsApp. I was asked to look at the situation directly
ClickUp and similar tools
Feature density overwhelmed the team from the first session
Significant setup required before any value was visible
Adoption lasted roughly a week in each case
The alternative: WhatsApp
No structure, no single source of truth for a given job
Files and approvals buried in scroll history
No connection to quoting, scheduling, or documentation
This wasn't unique to Veles
72%
of contractors overall report using some form of management software
38%
of small contractors, under 20 employees, use dedicated field tools
Small construction teams are consistently the most underserved segment of this market. Most available software is priced and designed for firms several times larger, which is part of why general-purpose tools like ClickUp felt like the wrong shape for a 20-person crew
The Product
Project management, AI-assisted quoting, an internal knowledge base, and scheduling, designed for a team that works mostly on-site
Chat Command Center. Ask questions, trigger actions, and surface insights in natural language. Every AI response shows its source data inline, reps can verify before acting



Field Research
I designed around the reality of the job site, not an ideal office workflow
I spoke with office staff, project leads, trades, and subcontractors, and observed how work actually moved through the company. Some conversations were closer to interviews; others happened while people were working, which made small friction points much easier to notice.
The strongest constraint was attention. Someone on site does not want to navigate a dense system to upload a photo, check a task, find an SOP, or answer a project question. That meant “simple” had to be defined by context, not by the number of features on a screen
Owner
"I'm spending hours on quotes instead of running the business"
Office staff
"I have to scroll back through a hundred messages just to
find a photo from Tuesday"
Subcontractor
"If the owner isn't available, the whole thing stops"
Research Synthesis
The recurring problems were less about project management and more about access, ownership, and repetition
Looking across interviews and observation, I grouped the operational pain into four areas: project context was scattered, important knowledge lived with specific people, quote creation created a bottleneck, and office/field users needed different levels of detail.
Those findings became the basis for prioritization. I did not try to replicate every capability of ClickUp or other project-management platforms; I focused on the workflows that repeatedly interrupted the team’s day
Scattered project context
Messages, files, deadlines, and updates were split across tools
Knowledge dependency
SOPs and answers often required asking the same people repeatedly
Quote bottleneck
Estimating knowledge was concentrated with the owner
Different environments
Office planning and on-site action required different information density
Product Scope
I defined the product around recurring jobs, not around a feature checklist
The core system became projects, tasks, project files, team chat, quotes, payment progress, SOPs, shared documents, and calendar visibility.
Just as important was what I did not prioritize. Advanced customization, complex dashboards, or broad automation would have increased setup and maintenance without solving the team’s main problems
Old workflow
Customer request → owner gathers information → calculates manually → creates quote → sends → revisions.
New workflow
Implementation
I used the working product as part of the design process
I designed the interface and system in Figma, then used Cursor and Figma-connected AI tooling to translate those designs into a functioning application. I was not manually authoring the production code; my role was to define the UI, behavior, states, and interaction details, then direct the implementation and continuously compare it against the design.
This changed the pace of iteration. Instead of waiting for a handoff cycle, I could see where a layout or interaction failed once it was clickable, test it with the team, and revise both the design and implementation in the same loop
Research Synthesis
The recurring problems were less about project management and more about access, ownership, and repetition
Looking across interviews and observation, I grouped the operational pain into four areas: project context was scattered, important knowledge lived with specific people, quote creation created a bottleneck, and office/field users needed different levels of detail.
Those findings became the basis for prioritization. I did not try to replicate every capability of ClickUp or other project-management platforms; I focused on the workflows that repeatedly interrupted the team’s day
Shared quoting
Other team members can create estimates without waiting for the owner
Faster preparation
The owner spends less time manually building quotes
Real usage
The office team is already managing active work inside the product
What I Learned
This experience sharpened my design thinking around AI-human collaboration
Internal tools benefit from restraint
The team had already tried more powerful software. The opportunity was not adding capability; it was identifying the smaller set of workflows people would actually use
Context changes what “simple” means
An interaction that feels straightforward at a desk can become cumbersome on a job site. Observing people in their working environment changed how I prioritized navigation, information density, and mobile actions
Implementation made iteration more immediate
AI-assisted development let me test design decisions in a functioning product instead of treating Figma as the endpoint. Seeing the interface used by the team exposed issues that were difficult to anticipate in static designs
·Files and approvals buried in scroll history