Geospatial site intelligence · App + Launch site

Can this site get power? Answered in hours.

Geoforge

2025–2026

Role: App product design, data-viz, front-end build
Timeline: 2025–2026
Platform: Web app · Public launch site (reports)
Stack: Figma · MapLibre · HTML · CSS · JS

“This whole thing is awesome, especially the blank state. Great across the board. Overall I'm really pleased with the direction of the design, it's bright and clean and clear.”

Client · Geoforge

Live product

geoforge.ai

Designed the app and the launch site, then vibe-coded the launch site to production myself. Final calls on look-and-feel are the client's.

Geoforge screens land for data centers. The hardest question on any site is whether it can get power, and that answer used to take a consultant weeks and thousands of dollars. Geoforge gives it back the same day, sourced. I designed the map workspace and the team side of the app, then designed and built the free launch site that delivers the report.

Overview

Power is the bottleneck for data centers, and checking it is expert work that takes weeks. Geoforge makes that check free and fast. Drop a location, get a report with a verdict on each part of the power picture. Two surfaces: the collaborative map app where teams keep their sites, and the public launch site that turns any pin into a shareable report.

Same day

vs weeks for a consultant power study

power criteria, each sourced

21,000+

points of interconnection modeled

to run a report, work email only

The problem

Screening a site for power means pulling together interconnection capacity, the queue, historical pricing, nearby generation and transmission, and time-to-power. Developers, landowners and brokers pay consultants weeks and thousands for that read. Most dead sites don't get killed until late, after money has already gone into them.

The data behind it all is dense and easy to argue with. The instinct is to boil it down to one overall score, but a single blended number just moves the argument somewhere else. The design problem was making a lot of technical grid data trustworthy to non-experts without hiding the parts that matter.

The solution

On the app side I designed the map workspace: a My Maps gallery where teams keep location datasets like grid capacity or candidate sites, plus everything around it (roles, invitations, settings, account). Deliberately light and quiet, so the maps and the data do the talking.

For the launch site the team made a call I still stand by: no blended score. Each report covers nine power criteria, and every one stands on its own with a verdict (pass, tier one, tier two or fail), the headline number and a sourced read behind it. Then I built the site myself: a self-contained page with live maps that anyone can run with a work email and share on its own link.

Screens and flows

Map workspace — The core map: data layers plotted over the grid, with a layers-and-legend panel alongside.

Choropleth map — value by county — The same data as a graduated-colour map: each county shaded by its value, with labels and a custom scheme.

Layers library — A team's datasets in one place — grid capacity, sites, land use — with background jobs running as data loads.

Data layer with an AI assistant — Editing a layer's rows, with an assistant that builds new data layers from a plain-language request.

Data layer — change history and revert — Every edit to a layer is logged and reversible: who changed what, by hand or via the assistant, with one-click undo.

Site diligence scorecard — Nine power criteria, each with its own tier verdict and headline figure — no single blended score.

Inbound site tracker — A pipeline of candidate sites with an advance-or-kill verdict, capacity and pricing for each.

Layer styling — Fill, size, colour scheme and opacity controls for how each dataset reads on the map.

Data mapping and geocoding — Matching uploaded addresses to map geometry before a layer renders.

Distance and drive-time — Measuring driving distance and time from a site to the nearest infrastructure.

Workspace — people and roles — Members, roles and invitations for a workspace, with guarded actions like removing a user.

What I designed

Map workspace

A My Maps gallery of location datasets with search, roles and sharing. A team's sites live in one place.

Nine-criteria power report

One verdict per criterion instead of one blended score. Each backed by a headline stat and a sourced read a non-expert can follow.

Trust through sourcing

Every figure traces back to grid and market records, modeled the way grid operators model it. The report holds up in a real conversation.

Designed and shipped

I designed the launch site and built it in code: a shareable web report with live maps, private to each account by default.

Keywords

product designgeospatialGISdata center site selectionenergy & grid datadata visualizationinformation designmap UIMapLibrePLG launch sitedesign-to-codereport designB2Bworkspace collaboration

Next project

Zest · Making AI adoption measurable