When Builders Show Up for Nonprofits

Most of my work lives in the hard conversations.

I spend a lot of time in rooms talking about what AI is doing wrong. The power imbalances. The privacy erosion. The governance gaps. The ways this technology, when it arrives without care or accountability, lands hardest on the people already carrying the heaviest loads.

So when Lautaro Cepeda invited me to judge the Builders Vault Social Services Hackathon in Victoria, I said yes before I finished reading the message.

I have always wanted to be part of a hackathon. But I'm not from tech, so I couldn't figure out how that would ever happen for me. I come from 20 years in the nonprofit sector. What I know is organizational systems, leadership under pressure, and the specific kind of exhaustion that comes from doing this work with not enough of anything.

That turned out to be exactly the background they needed.

The hackathon ran April 20 to 25 in Victoria. 23 participants. 5 days. 2 tracks, both built around problems that frontline social services organizations deal with.

Track 1: inter-organizational referral and care coordination, modeled on conversations with Cool Aid Society and lived-experience advocates from across Victoria's social services sector.

Track 2: food security delivery operations, grounded in the real operational structure of Fateh Care Charity.

Real problems. Real data. Real constraints. The people who designed the challenge understood that the gap between a brilliant prototype and something a frontline worker can actually use at 3pm on a Wednesday is enormous. They designed accordingly.

It's incredible what people can build in 5 days.

Mealflo tackled Track 2, the food security problem. Peter Salmon and Nicholas Miller built a prototype that takes food support requests, checks meal inventory and volunteer availability, and generates optimized delivery routes, with a phone-first driver interface and a live coordinator dashboard.

The problem it solves is unglamorous but relentless. Requests come in through texts, voicemails, and Instagram. Dietary restrictions and cold-chain needs live across spreadsheets. Every route gets pieced together manually.

A calmer operating rhythm. That's a solid goal for this sector.

Liaison took on Track 1, the referral and care coordination challenge. Dylan Whitbread and Muhammad Aarij built a morning-brief tool for caseworkers that surfaces likely duplicate client records across partner agencies, ranks them by confidence, and walks the caseworker through a one-click review.

The problem they solved is one I've watched nonprofit organizations navigate manually for years. A single person experiencing homelessness can exist as 3 or more separate records across different agencies, under slightly different name spellings, with no system connecting them. That fragmentation sits underneath every other coordination failure.

Liaison collapsed what used to take 52 minutes of phone calls and emails to 3. And it enforces consent and OCAP principles for Indigenous client data at the code level, not as a policy footnote.

Then there was Oflex, my favourite, and runner-up.

Have you ever tried to navigate a self-serve government portal? How did it make you feel? Was logging in straightforward? Imagine being unhoused and trying to navigate this.

MySelfServe is the BC government's portal for applying for income and disability assistance. For people who need it most, navigating it alone is genuinely hard.

Oflex built an AI-powered guide that walks people through the application process, grounded in direct conversations with Cool Aid, Our Place, and the Ministry of Social Development and Poverty Reduction. They spent their first days listening before they wrote a line of code.

There's something I appreciate about fixing broken systems that are right in front of us. Choosing to solve one real thing well, carefully, for people whose access to support could depend on it. That could be truly life-changing.

None of these prototypes are in production yet. That's the next question: what gets funded, and who follows up.

I've spent years concerned about the gap between the tech sector and the nonprofit sector, curious about how to bridge it.

The tech world moves fast, rewards disruption, and generally builds for people with reliable internet, stable housing, and time to learn new tools. The nonprofit sector moves slowly by necessity, is accountable to communities rather than shareholders, and operates on margins so thin that a failed software implementation can set an organization back years.

For a long time, those 2 worlds spoke different languages. The tech sector didn't always know what nonprofits actually needed. Nonprofits didn't always have the capacity to articulate it in a language that builders could act on.

What I saw at this hackathon was something different.

Builders showed up and asked good questions first. They spent the kickoff evening listening to sector practitioners before they wrote a single line of code. The challenge tracks were designed around what was operationally real.

AI gave those people the tools to move faster than would have been possible 5 years ago.

I talk a lot about the problems with AI. I will keep talking about them, because they are real and they matter and they don't go away when we look elsewhere.

But I also believe we are in a moment when the question of what we build, and who we build it for, is genuinely open. The people in that lecture hall at UVic on April 25th chose to point their skills at food insecurity and care coordination gaps in Victoria, BC.

Nobody made them do that. There was $500 in prize money on the table for each winner. That's not the reason people show up for 5 days and build something intentional.

Every team that demoed had a working prototype in 5 days. The 2 that won were the ones the rubric picked. But the others who built are part of what made the room work.

They showed up because the problem mattered to them.

That is heartening. Genuinely, in the old sense of the word. It puts something back.

The nonprofit sector has carried a lot for a long time with very little. It's good to see some of that generosity flowing the other way.

---

Sarah Downey is a Canada-based consultant who helps nonprofits adopt AI ethically through governance clarity, training, and AI policy development.