80% Faster in a Weekend: Building Emergency Response Software
The setup: It's February 2020. Colorado is about to face one of the coldest winters on record. People experiencing homelessness are lining up at City Hall in freezing temperatures, waiting 11 minutes per person just to fill out a 4-page paper form for emergency shelter.
The county's Severe Weather Action Plan could only process 60–80 people per night. Hundreds were left without help.
We had one weekend to fix it.
The Problem Nobody Saw Coming
When our Code for America fellowship team arrived in Adams County, we had what seemed like an obvious solution: build a smartphone app for people experiencing homelessness to request shelter directly.
Then we talked to actual people.
Only 10% had reliable internet access.
Our "innovative" app would exclude 90% of the people who needed help most. The cool technology we wanted to build would have been useless.
After 21 caseworker interviews, 10 stakeholder meetings, and 5 outreach visits, we learned something crucial: the bottleneck wasn't the people seeking help. It was the intake process.
The Weekend Sprint
Instead of building for end users, we built for caseworkers. Give them better tools, and they can help more people, faster.
We spent one weekend building a working prototype:
- Digital intake forms that replaced 4 pages of paper
- Tablet-ready interface for field work
- Automatic data storage for repeat visits
- Integration with hotel booking systems
Monday morning, we showed it to stakeholders. They approved it on the spot.
Real-World Constraints
Building software for emergency response means dealing with reality:
- No perfect internet: Caseworkers work in the field with spotty connectivity
- Freezing temperatures: Tablets in 10°F weather
- Emergency timelines: Cold weather doesn't wait for your sprint planning
- Vulnerable populations: Secure handling of sensitive data
We refined the system based on caseworker feedback and deployed to production within weeks. Right before the coldest part of winter hit.
The Results
80% faster intake: From 11 minutes to 2.5 minutes per person
15x capacity increase: 8,300+ nights of shelter provided (vs. 500 the previous year)
560+ people served during the 2020–2021 cold weather season
422 vouchers distributed in February 2021 alone—a record month
As one caseworker, Nubia Saenz, put it:
"[The app] has allowed me to go through the line faster, and [our clients'] anxiety has decreased because they don't have to fill out so much paperwork, and it has allowed them to speak up more on their needs."
What This Actually Teaches Us
1. User research kills bad ideas early
The smartphone app would have been a disaster. We only found the real problem by talking to people doing the work.
2. Perfect is the enemy of deployed
A weekend prototype beat months of planning. In an emergency, working software matters more than elegant code.
3. Build for the bottleneck
We wanted to help people experiencing homelessness. The fastest way to do that? Help the caseworkers who serve them.
4. Constraints force better solutions
No internet, field conditions, emergency timelines—these constraints didn't make the project harder. They made the problem clearer.
The Takeaway
Technology doesn't solve homelessness. People do.
But the right technology, built fast and deployed in the real world, can free up those people to focus on what matters: getting someone shelter before the temperature drops.
Sometimes that's worth more than a perfect sprint plan.
Want the full technical breakdown? Read the complete SWAPP case study for details on the engineering, integrations, and lessons learned.
Building something under pressure? Let's talk about how we can help you ship working software fast.