SellCRE turns one property address into a finished marketing package for commercial real estate brokers.
Before a property can be sold, a broker builds a package around it: an offering memorandum with property details, financials, maps and comparables, plus brochures, flyers, ads and emails.
I designed SellCRE from the ground up: the flows, the structure, every screen and the templates. As the product grew, I grew the design team around it and led five designers.
Design process
The problem
To market a single property, brokers copied the same numbers from reports into spreadsheets, documents and design files. Every hand-off was a chance for errors and delays, and every package looked different.
How might we
help a broker go from one property address to client-ready marketing, with AI doing the heavy lifting and the broker staying in control?
Who we designed for
Proto-personas built from what the team knew about our broker users. They kept every decision tied to a real job to be done.
Primary user
Senior broker at a commercial brokerage
Wins mandates and needs listings on the market fast. Spends time with clients, not in spreadsheets.
Goals
Frustrations
Secondary user
Supports several brokers on a team
Turns brokers’ notes and numbers into polished packages, and owns brand consistency.
Goals
Frustrations
User flow
I mapped the whole journey before drawing a single screen. Every step falls into one of three stages, and the stage is always visible at the top of the app.
01 Start
02 Upload
03 Review
04 Publish
Information architecture
An OM covers a lot of ground. I grouped every field into five review tabs, each with a short list of sub-steps, so brokers only ever see one small form at a time.
Wireframes
Low-fidelity layouts of the four key screens. They let us agree on hierarchy and navigation before any visual design.
Visual system
Brokers handle multi-million-dollar properties. The UI stays light and quiet so photos and numbers stand out, with one confident purple for actions and progress.
Poppins · semibold, medium, regular
Property Summary
Selling points
Situated just 30 minutes from bustling Atlanta.
Components
Following one listing through the live product, from first login to a client-ready offering memorandum.
01 Onboarding
A short two-step onboarding captures the broker’s profile, company logo and brand colours. Every template then uses them automatically.
02 Start a package
Templates and fields adapt to the asset type. Address autocomplete prevents typos, and the data starts loading straight away.
03 Auto-filled data
Property information, features and sale comparables are pulled in automatically. Brokers check and correct instead of typing.
04 AI data extraction
Upload a CoStar report, rent roll or cash-flow sheet. AI extracts the details and fills the forms, and every value stays editable.
05 Review wizard
Brokers choose whether AI writes the summary or they write it themselves. Selling points are short, structured and always editable.
06 Underwriting
Unit mix, income, expenses and five-year projections in a familiar grid, with totals calculated automatically.
07 OM editor
Every page of the offering memorandum in one list. Brokers reorder pages, swap templates and edit text and images right on the page.
08 Property dashboard
Photo, wizard progress and every generated material for a property in one view, ready to open, download or share.
09 Marketing website
A homepage and feature pages for AI property data, AI financial data, OM creation and custom maps, written and laid out for brokers.
What I learned
Making AI a choice, and keeping every word editable, made brokers comfortable using it.
The upload, review, publish model did more for usability than any single screen.
Building the front end myself meant fewer gaps between the design and the shipped product.
Next project
