As a lawyer, I was a fish out of water. The pace, the risk-aversion, the gatekeeping — I felt constantly ill at ease among people who clearly felt right at home. My stint as a consultant ended in me being let go, and left me with the same existential question: could this really be it?

So I trusted my instinct and wound my way into startups. In a short period of time, I was running a successful venture-backed parking startup. It gave me the speed, ownership and creativity I’d been looking for, but a question about meaning still lingered.

As company CEO, I traveled to San Francisco often: meeting people, raising money, attending events. But on one occasion, I accidentally wandered into the Tenderloin, the city’s notorious downtown neighborhood.

It was Christmas Day, bitingly cold, and I chanced upon an open-air church service. On paper, that might sound quaint. In reality, it was the most desolate scene I’ve encountered. People not wearing shoes, human unmentionables littering the pavement.

This scene would have been shocking in any city on earth. But I wasn’t in just any city — I was in San Francisco, where one in every 14,000 residents is a billionaire. I tried to square this in my head. When I couldn’t, I realized it was time for another job change — this time, to build a very different kind of business.

The problems we chose to solve instead

In the decade since that trip to San Francisco, inequality has widened. Meanwhile, the technology industry has done next to nothing to address the most urgent of human problems.

What it has done — both brilliantly and expensively — is solve the problems of the people who didn’t have many in the first place. Billions of dollars’ worth of this generation’s finest engineering talent has helped make takeouts arrive quicker while applying the cutting-edge machine learning to serve ads faster and more profitably. I don’t sneer at any of these developments — I ran a parking app, after all, and I’m glad that my food arrives quickly and conveniently, but it raises a question that we don’t ask ourselves enough. Why does so much effort go towards making our already comfortable lives more convenient, all while so little goes towards the people who are most underserved?

What we were ignoring

While Silicon Valley was making some lives more convenient, many lives got a lot more difficult. In the fiscal year of 2025, SNAP served an average of 42.1 million people every single month. That amounts to 12.3% of the U.S. population — roughly one in eight people in the richest country in history — needing help to buy their food.

Now consider how that level of support is administered. In the fiscal year of 2025 the USDA reported a national payment error rate of 10.62 percent, worth $10.1 billion in improper payments. That’s more than one dollar in every ten. When broken down, around 87 percent of that was overpayment, but that leaves roughly 13 per cent as underpayment. That translates as eligible families not getting the support that they are owed.

The USDA is explicit that this is not fraud — these are errors, a truth that is constantly misreported in the media. So, what is causing so many errors? Overwhelmingly, they are the result of inordinate amounts of paperwork, outdated systems, handwritten notes, and sheer human exhaustion.

Clearly, this is a systems problem. And what’s brilliant at fixing systems problems? Silicon Valley’s favorite friend: software.

The bit the spreadsheets miss

Behind every one of those error rates are people sitting on either side of a desk. One is seeking support, the other is administering it.

A federal study by the Administration for Children and Families published in 2025 found that child welfare caseworkers spend an average of 4.3 hours a day on paperwork and documentation; over half of a standard eight-hour shift.

That translates as recruiting people whose entire vocation is centered on sitting with a family in crisis, and then forcing them to spend two thirds of their working life filling out forms. We then act surprised when they grow tired, burn out and leave the profession altogether; national average annual caseworker turnover in U.S. child welfare runs at around 30% (roughly 3 in 10). And the person who needed them often slips through a crack that a better-designed system would have helped prevent.

Ten years ago, soon after that defining trip to San Francisco, I gave up on building convenience technology and started building technology for social services instead. Now, I create AI tools for people on the front line and for the vulnerable people that they serve, both of whom the technology sector has widely neglected. For the last decade, my mission has been simple: to put technology — and now AI — at the service of society. Today we’re operating across five countries and in some of the largest cities in the world, including in California.

Our first U.S. partner was Fairfax County, Virginia, and today agencies in a dozen U.S. states are cutting the admin burden in half. But this is about so much more than efficiency. This is about supporting thousands of frontline staff in a sector where burnout is commonplace, ensuring that every conversation is documented thoroughly, and ultimately making sure that more vulnerable people get the support they need.

Who technology forgot

These days most of my time goes on one thing: persuading people who have technical skill to point it at something that matters. It’s the conversation I wish I’d had earlier in my career.

There was a time in the early days of the Internet when tech saw itself as being in service of something bigger than itself. Somewhere along the way, that gave way to a race for profit. By measure of money, Silicon Valley has never done better. But it won’t rediscover purpose by solving the inconveniences of people who are already comfortable.

The hardest problems facing humanity are unglamorous, badly funded, stifled by bureaucracy and full of forms. But they are also right there, in plain sight, on a freezing pavement just round the corner from the money.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.