Building AIHMA From the Ground Up

AIHMA came to us needing a site that could hold the weight of what they’re building: the first institute dedicated to AI governance within mobility and immigration. No off-the-shelf nonprofit template was going to carry that. It needed to be built from scratch.

Who AIHMA is

AIHMA — the AI Institute for Human Migration and Arrival — sits at an intersection most organizations haven’t touched yet: accelerating human migration on one side, and lagging AI accountability on the other. Their work spans workforce mobility, refugee resettlement, climate displacement, and the responsible governance of the AI systems increasingly making decisions about people’s lives.

Two programs anchor what they do. Arrival Intelligence is applied research aimed at helping newcomers arrive, settle, work, and thrive faster in new workplaces and communities. The AI Governance Accelerator gives nonprofits serving mobile populations the governance playbooks and adoption plans they need so AI drives human and organizational benefit, not harm.

Built from scratch, accelerated by AI

Nothing on the AIHMA site started from a theme. Every section came out of working closely with AIHMA’s founder — sitting with their vision for the site, going back and forth on how the research framing, the program breakdowns, and the governance language should actually read, until the site matched what they’d pictured rather than what a template assumed a nonprofit should look like.

What made that possible on a real timeline was leaning on AI as a genuine lift in the build itself: scaffolding markup faster, iterating on layout options in minutes instead of hours, catching inconsistencies before they shipped.

A modular, conditional page shell

The page shell was built to bend, not to be rebuilt. Sections are modular — self-contained blocks that can be reordered, swapped, or dropped without dragging the rest of the page’s layout logic along with them. Underneath that, the lines that render them are conditional: a section only renders its full treatment when it actually has something to show, and falls back cleanly when it doesn’t, instead of leaving a placeholder-shaped hole in the page.

That shape was an engineering call, not something AI handed us by default. AIHMA is a young institute — their programs, research, and public-facing content are going to expand, and some sections won’t have anything to show yet. So the architecture was built around that from the start: clear boundaries between components, state that decides what renders instead of markup that assumes content is always there. That’s the kind of decision that comes from thinking through how a system needs to hold up over time, not from following whatever a tool suggests. AI can write the lines. It doesn’t decide where the seams go.

Where the human touch came in

The speed came from AI. The feel of the site didn’t.

Working alongside our designer on this build is what kept AIHMA’s site from reading like a fast, generic output. Layout decisions, type pairing, the restraint to leave space alone instead of filling it: that’s design judgment, not something a prompt produces on its own. AI moved the build forward. A person made the calls about what the organization actually needed to feel like.

That’s the model, not a one-off. We’re not interested in spitting out AI-generated websites and calling it a service. We believe in the collaboration — emerging technology doing the heavy lifting on speed and iteration, human judgment doing the work of making something feel considered, specific, and actually built for the organization in front of us.

Who we’re looking to work with next

AIHMA is the kind of project we want more of: an organization doing work that matters, open to a development process that doesn’t look like the last ten years of agency defaults, and looking for either a first real online presence or a refresh that finally matches what they’re building.

If that’s you — and you’re not willing to trade human connection for speed — we should talk.