Row makes it easy for talent and capital to find the teams doing the world's most important work.
AI went from research project to global infrastructure in about three years, and the upside is already real: faster drug discovery, tutoring that adapts to one kid instead of thirty, code and analysis that used to take a team a week done in an afternoon. Nobody serious wants to slow that down for its own sake.
The same speed cuts the other way. The systems keep getting more capable before our institutions can fully digest them. The U.S. has leaned on voluntary commitments while individual states pass their own laws, and labs test their own systems and publish the results, which is better than nothing but not the independent oversight we need.
The technology moves fast. The rules don't keep up. That gap is what this arena is about: capturing the upside without losing control of the downside. Everything below is aimed at closing it.
More solar came online globally last year than every other energy source combined, and clean electricity is now the cheapest electricity in most of the world. Building the clean power itself is going well, but wiring it into the grid gets less attention. Our infrastructure is built on oil and gas, and transitioning will take time and tough choices.
Reliable and affordable are lagging behind. Most of the grid carrying that power was built for a different era. Policy support has also gotten less predictable, which makes the multi-year bets this work requires harder to place.
Everything below is aimed at moving clean, reliable, and affordable closer together instead of trading one for another.
We focus on a few critical mechanics: elections that produce results people accept, checks that keep any single branch from running unchecked, a Congress that can get legislation passed, and reining in money in politics.
Some of the work is protecting dynamics that already function: courts still ruling, agencies still enforcing the law, elections still getting certified.
The rest is building versions of those mechanics that hold up better under pressure than the current ones do.
The larger project is keeping a two-hundred-year-old system flexible enough to bend without breaking. Rules written for a slower era have to adapt to a faster one, and stay trustworthy along the way.
The downstream effects of a disconnected, squeezed culture show up everywhere: polarization, mental health strain, falling birth rates, people who can't quite say why they're not thriving. I picked the dynamics on this page because they felt load-bearing: meaning if we solve one, the pressure on the rest eases up.
None of this has a clean legislative fix. Nobody passes a bill that makes people feel less alone or more hopeful about their kids' future. The people working this arena are building better versions of community, care, and meaning, brick by brick.
Everything below is that shortlist: the handful of things that, if they get better, make the rest of a good life more possible.
We've lost the infrastructure for a shared, baseline understanding of the world. A healthy information environment gives people reporting they can trust: local coverage of their own communities, platforms that surface what's true instead of just what's engaging, and enough transparency to tell a credible source from a manufactured one.
Most of that infrastructure doesn't exist yet. It's on us to rebuild it.
The organizations here are working all three. Most are working one.
The information layer that holds democracy together is being built and demolished at the same time.
Government agencies, universities, and big foundations were built for a slower, more predictable century, and the pace of change is outrunning their ability to keep up.
That's the problem underneath all the others: how we solve problems matters as much as what we're solving. The people inside these institutions are usually excellent. What limits them is the shape of the institution itself, how it's funded, how fast it can move, and what it's legally allowed to do. The work here is inventing better shapes: new legal structures, new funding models, new ways to move money and talent, instead of defaulting to the traditional nonprofit playbook every time.
Anthropic wrote the frontier safety framework first and the rest of the industry copied the shape of it. The Responsible Scaling Policy landed in September 2023, ahead of OpenAI's Preparedness Framework and Google DeepMind's Frontier Safety Framework, and METR now tracks a dozen companies running some version of the same idea. Version 3.0, effective February 2026, added public Risk Reports: quantified, externally reviewed disclosures of how dangerous the company believes its own models are. They also fund the unglamorous half of the field, interpretability and alignment science, where results take years and never ship as a feature. This is the lab that publishes what it's afraid of.
Standard Setter. Writes the safety commitments before regulators do, publishes them in full, and eats the cost of going first. Competitors get a template and governments get something concrete to point at.
Most climate money funds research or funds finished projects. Elemental built the instrument for the gap in between. Its D-SAFE, developed with Wilson Sonsini and open-sourced as a free template in 2025, hands startups non-dilutive money for permitting and engineering work before a project is bankable. Roughly $7 million in D-SAFEs has unlocked more than $70 million behind it. Across seventeen years and 160-plus companies, every Elemental dollar has pulled in about $110 more. In May 2026 they got Amazon, Google, Meta and Microsoft to co-fund startup pilots inside their own data centers.
First Project Financier. Pays for the years between a working prototype and a financed first plant, then publishes the paperwork so other investors can copy the structure without paying the legal bill.
Most data-for-good efforts fund a project. data.org funds a labor market. It launched in 2020 on a $50 million commitment from the Mastercard Center for Inclusive Growth and The Rockefeller Foundation, built on one premise: the binding constraint on data solving public problems is that nonprofits and governments cannot hire anyone who can build the solutions. The Capacity Accelerator Network now runs five regional hubs across the United States, India, Africa, Latin America and Asia Pacific with more than 85 partners. Nearly 100,000 people trained as of early 2025, against a target of one million by 2032.
Workforce Builder. Everyone else in data-for-good builds tools. data.org builds the people who use them and strengthens the organizations that hire them, so the tools have somewhere to land.
Plenty of people criticize AI products. Very few can explain to a federal judge how one was actually built. CHT worked as expert consultant to plaintiffs' co-counsel in Garcia v. Character Technologies, briefing the legal team on the product's mechanics and researching whether the psychological harms were systemic rather than isolated. In May 2025 the court let the case proceed and accepted that an AI system can be a product under liability law, which is the doorway every design-accountability case now runs through. Google and Character.AI settled in January 2026.
Harm Translator. Converts inside knowledge of how products get built into forms institutions can use: legal filings, policy testimony, and language ordinary people can reach for to describe what a design did to them.
Most people fighting platform lock-in pick one lane, the protocol or the law. Project Liberty runs both, and the halves compound. Its Labs arm shipped DSNP and Frequency, a live data layer the organization says now secures records for more than 16 million people. Its Institute then provided the technical assistance behind Utah's Digital Choice Act, the first US law requiring social platforms to let users take their data and their connections elsewhere. A working protocol, plus a statute that creates demand for one. Frank McCourt has put $100 million behind it and an alliance of 175-plus organizations behind that.
Standard Setter. Writes the technical standard and the law that requires it at the same time, so portability has a working protocol to point at and a legal reason for platforms to adopt it.
Axios proved a thesis the industry had written off, that readers are busy and smart at the same time. Smart Brevity, stories under 300 words with the stakes stated outright, got treated as a gimmick until it sold for $525 million. Local is the harder bet. While the chains stripped newsrooms, Axios put more than 100 reporters into 41 metros, aiming at suburbs and news deserts identified through Northwestern's research, with OpenAI helping fund the expansion. Local now brings in tens of millions with several profitable markets. Nobody else is opening local newsrooms at this rate.
Format Proof. Shows that a news format built around the reader's time can carry a national business, and then travel into local markets that other operators had already written off.
Emerson Collective is built as an LLC instead of a private foundation, which frees it to do things a foundation legally cannot. IRS rules cap a foundation at 20 percent of a company's voting stock, and Emerson holds a majority of The Atlantic. It writes venture equity checks and funds open political advocacy. It also builds and staffs its own organizations rather than only granting to them, including XQ Institute, Chicago CRED and E Pluribus Unum. When a vehicle outgrows the house it spins out with its own balance sheet, the way Waverley Street did for climate in 2021.
Full-Stack Funder. Holds grants, equity, outright ownership and political advocacy on a single balance sheet, then picks the instrument a problem actually needs instead of the one a charter permits.
Helena runs two entities under one brand, a 501(c)(3) foundation and Helena Special Investments. The stated process picks a problem first and then assigns it a vehicle, which might be a grant, a company, an equity stake, a policy campaign, or a four-day deliberative poll. Membership carries no fee and no application, and nobody can buy in. A conventional foundation has to find an existing grantee, fit the work to its program areas, and cannot take equity in the answer. Helena can start the thing and own part of it.
Vehicle-Agnostic Operator. Starts from the problem instead of a program area, then picks the legal form that can carry the work: a nonprofit, an LLC, a poll, a bill, or an equity stake.
Frontier, a public benefit LLC wholly owned by Stripe, is a purchase order written years before the product exists. A grantmaker funds work that has already been scoped. Frontier signs eight- to ten-year offtake contracts, some running to 2040, that a supplier can carry into a bank as collateral. That converts philanthropy-shaped money into project finance, which is the real bottleneck for a first-of-a-kind plant. The structure travels: pool unrelated companies' commitments, run technical diligence once instead of ten times, buy on their behalf. Any industry with an expensive unbuilt supply chain could copy it. Almost none have.
Demand Guarantee. Turns scattered corporate climate budgets into one credible promise to buy, which is the thing that lets an unbuilt carbon removal plant raise money today.
Most philanthropy moves money toward problems. This moves people. The school treats elite career choice as the scarce input and competes for it directly, running an application funnel built to look like McKinsey's and aimed at the same candidates. Fellows get salaried and embedded inside existing host organizations rather than pushed to found new ones, so a placement strengthens a group already doing the work. And the school picks its problems first, tobacco and factory farming and tax fairness, then recruits against them. A funder cannot do that. An employer-shaped institution can.
Talent Redirect. Intercepts high-skill people at the moment they are choosing a career and hands them a funded, low-risk path into one specific unglamorous problem.
If you want to know how the economy is doing, you look at the S&P 500. If you want to know who's playing good ball, you look at a league's standings. Restaurants have Michelin, movies have the box office.
The biggest problems facing humanity have no such thing.
It shouldn't be hard to name the five teams doing the best work on AI safety. Or the grid. Or mental health. Or getting money out of politics.
The information is out there, scattered across a thousand reports, newsletters, pitch decks, and group chats, but nobody has put it in one place where a normal person can engage with it.
I think this is a huge miss. Partly because it keeps these problems and teams from being broadly legible, and partly because engaging with these organizations tends to inspire the kind of optimism and creativity we need more of.
The inaugural row99 has 99 seats across six problem categories.
Each seat goes to one team doing the best work on one of humanity's biggest problems. Right now that's AI, climate, democracy, culture, information, and what I'm calling "method," teams changing how we solve problems.
Unlike other lists of nonprofits or innovative companies, the row99 is sector agnostic. If a team is doing great work, they're on the list.
As the index grows, so does our picture of a field. "The people working on democracy" becomes a thing you can see. You can navigate it, compare teams, and argue about who should be in there. That argument is the sign it's working, btw.
A foundation and an investment arm under one roof, picking the problem first and then choosing the vehicle: a grant, a bill, an equity stake, a citizens' assembly.
To start with, there's no committee and no application. I read the field, deeply research the players, look at what the work has produced, and make my selections. For every seat, I do a quick 90-second explainer on who they are, why they're distinctive, and the outcomes they're driving for humanity.
Feedback and criticism are always welcome.
For humanity to get its sh*t together on its biggest challenges, we need a shared space to understand the problems and key players. Row hopes to build just that.
99 seats, filled one defense at a time. Here's the bar, so you can hold me to it.
Every seat attaches to at least one thing that has to become true. If I can't name the line an org moves, they don't get a seat — however famous they are.
A specific, dated thing they did. A mission statement is a plan; a receipt is proof. Every profile carries at least one.
If they stopped tomorrow, someone working this problem would feel it within a year. That's the difference between leading work and adjacent work.
Companies, nonprofits, funders, media — the grid holds all of them, because the problems don't care about corporate structure and the leading work is wherever it is.
Every seat comes with my reasoning and my caveat, in writing, with my name on it. When I get one wrong, the correction ships with the same volume as the pick.