Assembli AI
Create construction estimates in minutes.
Role
Design Consultant
Tools
Figma, Notion, ChatGPT
Year
2025
Estimating construction costs is slow, error-prone, and stressful—especially for small to mid-sized builders. Assembli set out to fix that.
By combining AI with real-time data, we help contractors generate accurate estimates in minutes—not days.
The MVP needed to:
Spark investor confidence
Attract early adopters
Prove the product works
Lay a scalable design foundation
Deliver fast, despite tight time and resources
We shipped a lean, testable MVP in under 8 weeks that gave the team what they needed:
+75%
Faster Bid Turnarounds
-52%
Fewer Budget Overruns
+88%
MVP
Used in Funding Pitches |
We moved fast, but didn’t skip the why.
Interviews with contractors revealed fears: inaccurate AI, complex flows, and fear of "getting it wrong."
Jobs To Be Done clarified core needs: speed, accuracy, and simplicity
Used ChatGPT to rapidly prototype UX writing and edge cases
The insight that stuck:
“If I can’t trust the numbers, it’s useless—even if it’s fast.”
From spreadsheet stress to step-by-step clarity
Broke job setup into digestible steps
Added inline helpers to ease uncertainty
Synced UI with backend logic to reduce timeouts + data loss
Start fast, then make it yours
Introduced smart presets for labor, materials, and measurements
Made customization optional—not required—to get going
Result: Faster setup so builders could get right to estimating
One view, multiple jobs—no chaos
Created a dashboard-style layout to reduce context switching
Introduced smart summaries for at-a-glance control
Build once, scale infinitely
Built modular components from Day 1
Enabled design–dev handoff in hours, not days
Set up token-based theming for future flexibility
Working closely with the founding team and engineers, we prioritized speed without sacrificing UX. I hosted async design reviews in Notion and ran quick huddles to unblock devs
One key moment:
We discovered that users would have a long load time when the system started to analyze the documents. Instead of a long pre-loader we used clean animation to make the experience seem seamless so when they reached the end the documents would be analyzed
