A construction permit that took four months now takes three minutes

A construction permit used to mean more than forty process steps, a roster of paid consultants, and a wait of three or four months. BrainCo, an AI company Jared Kushner co-founded, found a government whose records were already digitized and, after enough trust had been built, was allowed to work through that stack. The original bet was modest: if a model could hit 90 percent accuracy on a single step, that would count as a win. Instead the team cleared that bar on more than thirty-five steps, and approval time fell to three minutes.

The person telling that story is Jared Kushner: former senior White House adviser, a central operator on the Abraham Accords, and now founder of the Miami investment firm Affinity Partners. In September 2025 he sat down with investors Sarah Guo and Elad Gil on their podcast *No Priors*. The conversation ranged across three files he is carrying at once — a global investing book, a new AI venture, and the geopolitics that now run through both.

He treats those files as one problem rather than three careers. Affinity, he says, should be an outcome-determinative partner to the best companies, putting capital in so the advice is not a consulting invoice. BrainCo is the implementation layer he could not find for his own portfolio and for Gulf partners. And the Middle East, in his telling, is where Gulf capital, Israeli technology, and American models are already colliding.

After leaving government, Kushner studied the firms around him and decided the world did not need another couple-billion-dollar private equity shop. What friends who invest actually needed, he concluded, was a partner who could solve messy problems and then put skin in the game. Affinity's pitch is to find the best companies, map their hardest challenges and biggest openings, invest alongside them, and then use a large, trusted network to bring customers and opportunities. Consultants invoice; Affinity writes a check.

Most operators, he argues, become excellent by obsessing over one industry or one geography, which also blinds them to how the same business is done in India, Europe, or the Gulf. Gil offers a historical rhyme: Yuri Milner had built a Russian social network, saw unused monetization in Facebook, and paid a valuation that looked insane when the company was worth about $5 billion. That cross-border eye, Gil notes, is the gap between a few billion and a trillion-ish market cap. Affinity has used a similar generalist's license on four insurance investments it likes as long-term compounders, and on Brazilian food: a large Burger King position, then Starbucks and Subway out of bankruptcy with Mubadala. The next question on those books is how AI changes the margin and the growth.

What government taught him about underwriting is a cocktail. At their best, governments set policy; the private sector still has to bring technical expertise, project management, capital, and local navigation. Three of those four, he says, is a recipe for disaster. Countries, unlike listed stocks, are not repriced every day — there are a couple of hundred of them, and they are inefficient — so the first screen is whether leadership is creating tailwinds a private investor can ride. He likens it to surfing: a mediocre surfer on a terrible wave looks worse than that same surfer on a great one. Albania is the example friends called crazy three years ago. He saw location inside Europe, a prime minister, Edi Rama, who is technically a socialist and still pro-business, and an artist-by-training with taste for the country's public face. A friend who visited this summer called back asking whether he could buy a house and join the projects.

Singapore, in his White House years, was the first desk check on any ugly new problem: what did they do there, because nothing under the sun was actually new. After government he had to interrogate a possible recency bias, then still concluded that Saudi Arabia, Qatar, and the UAE were on the rise, with toolkits of capital and industry that can accelerate innovation. Latin America is the other book, especially Mexico as U.S. companies hesitate to keep supply chains pinned on Asia, if cartels, immigration, and trade can be set on fairer terms. The return that made the geopolitics concrete was Phoenix Holdings, Israel's largest financial institution, with about $160 billion under management. Affinity became its largest shareholder in July 2024 at about $37 a share, just before a northern war with Hezbollah that made the investment committee miserable. Kushner argued the price was advantageous and that Gulf limited partners being more vested in Israel's economy was itself a feature. They bought 4.9 percent, then another 4.9 percent once regulators allowed it, and used the rise in between; a year later the stock was almost $120, nearly a nine-times return. Phoenix is now looking at infrastructure in Saudi Arabia and the UAE. "When you get an Arab and Israeli together and they start talking about a business opportunity, they forget about everything else."

Affinity went looking for serious AI help for its portfolio and for Gulf partners and found a market of ineffective tools, rip-offs, and happy talk. The same week, partner Luis was in the Middle East walking sovereigns through AI roadmaps and heard the identical complaint. They took the frustration to Gil, who is also a BrainCo co-founder; his reply was to start a company. Kushner, Gil, Eric Wu, and Luis began in January 2024. Dan and Marcia later joined as part of the founding team, and Clemens came on as CEO. The firm now has more than forty people and is hiring against demand rather than against a white space.

Gil is careful about the category: this is a platform company, not a body shop. A shared infrastructure layer is reused across a client's divisions; applications built on top get resold into other giants. Kushner had watched, around Thrive, how the best technologists build tools while large corporations hesitate to take risk on emerging tech. In government he had watched world-class CEOs spend their days keeping their jobs and running sprawling businesses, hearing about AI without the people or the access to implement it. Silicon Valley, he says, is superb at solving problems and less good at naming which problems to solve. BrainCo is supposed to sit on that seam.

The live cases already span healthcare (patient experience, throughput, cost), hospitality (reservation clusters and faster booking), and insurance claims. The construction-permit project is the one that delighted a former real-estate operator. Digitizing the workflow exposed that humans had designed a process with more than forty steps, of which more than ten could simply be deleted. A later model update also unlocked the conversion of spatial plans into something a model could score. Consultants who charged for busywork lose a market; developers build and pay tax sooner; governments lose a pocket of opaque judgment that Kushner says is where graft lives. Customers of this size do not buy on a pitch. The team did a lot of work at risk to prove technical depth, skipping the usual startup path of SMB, then mid-market, then enterprise seven years later. BrainCo, he says, started at year seven.

A good early customer, in his screen, has a complex problem with obvious efficiency on the table, the right data to build against, and a command-and-control structure willing to force a try. Miss the third and the first two do not save you, because people fear change. "Change is like heaven. Everyone wants to go there, but nobody wants to die." Some of the most resistant operators, once they see the work getting easier, become the loudest champions. Gil adds that the real blocker in enterprises is usually people, process, and organizational redesign, not the model. The company's near-term problem is not demand: it is keeping up, picking the right projects, and hiring without diluting the bar, while early clients expand scope across more divisions.

Kushner's first tour in the White House already had an industrial-policy edge. At President Trump's direction he reached out to TSMC and pushed the company toward U.S. fabrication; Arizona, he says, is the foundation of a larger build. He likes the current administration's AI roadmap and credits David Sacks with running it. The plan, as he boils it down, is two moves: assemble the tools to compete, which mostly means accelerating energy, and do not smother the technology before anyone knows what to regulate.

Energy is not a switch. Permitting has to get faster, and he thinks American private capital will fund data centers if the government sets the environment — including a nuclear cycle he would like to see become far more efficient. The graph that made him optimistic came from a U.S. battery company he looked at two years ago and did not buy: about 1,300 gigawatts of existing capacity, and about 2,000 gigawatts sitting in the FERC queue. Money and will are present; the queue is the choke. He contrasts that with Europe, where early rules are, in his view, why little is happening on AI — convenient for America, and a warning not to regulate from worst-case hypotheticals until the worst cases are clearer. The other half of the pitch is talent: pull the best people to the United States so the technology ships under American values.

The Gulf has been an AI player longer than the current boom made obvious. In 2018, while working on the Abraham Accords, UAE national security adviser Sheikh Tahnoun would not take a meeting unless Kushner stayed afterward to see what he was building in AI. Kushner was unimpressed at the time and later watched Mohammed bin Salman, Mohammed bin Zayed, and Qatar's emir and prime minister, Sheikh Mohammed, all put serious chips on the table. They talk, they compete, and they have different strengths: Saudi Arabia has the larger domestic market; the UAE wants to be a large investor and an implementer. Affinity's job in the middle is to connect that capital to leading U.S. AI companies rather than Asian ones. He thinks the region becomes not only an investor but a sandbox — companies, governments, and data that can birth tools, applications, and even new firms for export.

Kushner says he got the Middle East almost by accident. After the 2016 campaign, he and Ivanka were still deciding whether to move to Washington when a family-company publicist called: the *New York Times* had been told he was going, and that he would work on Middle East peace. The source was Donald Trump, giving an interview. The next day's front page asked who Jared Kushner was and what he knew about the region. His own answer, he tells Guo and Gil, was nothing; nobody there knew him either. If that was the assignment, he spent the first six months to a year asking some version of the same question: if you were us, what would you do?

He preferred failure autopsies to the resident experts. "Only in Washington," he found, could someone fail at a problem for a decade or two and still be introduced as the person to ask. The inbox he inherited included an ISIS caliphate the size of Ohio, ruling about eight million people and beheading journalists, plus a Gulf that felt betrayed by the Iran deal and had started learning Chinese on the theory that America was no longer a reliable partner. His method was to lock an end state with every party first. Once the destination is shared, he says, you are no longer sitting across the table; you are iterating on the path. The Abraham Accords, he jokes, succeeded on plan C, "only because we went through the alphabet like three times and failed at every letter along the way." In Silicon Valley, taking on an impossible task is celebrated; in politics, he argues, people avoid hard problems because failure is likely, so they do not try.

Conventional wisdom said Arabs would not come together with Israel until the Palestinian file was solved — a riddle he thought had been designed not to close. He pushed people to look forward, on security (Iran as a shared menace) and on the match between Israel's technology scene and Gulf AI ambition. Israel-UAE took three and a half years. Bahrain, Sudan, and Morocco followed; Kosovo wanted in alongside a Serbia deal; on January 6, 2021, they closed the GCC dispute among Qatar, Saudi Arabia, the UAE, Egypt, and Bahrain. From August 2020 through that last agreement, the pace was a burst. When Trump left, Kushner says a Saudi deal could have been done in three to six months. The next administration, in his account, reversed course because the policies were Trump's, then rediscovered the file after China helped broker between Saudi Arabia and the Houthis.

He takes seriously the analysis that Hamas's October 7 attack was meant, in part, to wreck a forming Israel-Saudi normalization. Iran, he argues, needs the Jerusalem narrative: the Quds Force is named for Jerusalem, and the revolutionary government uses the cause to justify itself. After the war he lists a changed board — Hezbollah's leadership gutted, Iran weaker and its nuclear program set back, Houthis under pressure, Hamas and Gaza largely destroyed — and says the remaining work is to convert military wins into political ones, which requires a Gaza settlement and the hostages home. Syria and Lebanon have new leadership; with Iranian influence reduced, he sees a real chance at stability. Trump and envoy Steve Witkoff are pushing the next phase of the Accords, with a long queue that still includes Indonesia and Pakistan once Gaza is resolved.

Guo and Gil note that unusual numbers of accomplished technologists are now walking into Washington. Kushner names Airbnb co-founder Joe Gebbia as U.S. chief design officer and Antonio Gracias, whose work on Social Security payments he says has already saved more than $10 billion a year. He is satisfied by what Elon Musk started with DOGE: people being hired at the third and fourth levels, he claims, are stronger than those willing to take top jobs in the first Trump term, and private-sector operators are renegotiating contracts with what they describe as easy, billion-here-billion-there concessions. The government, in this telling, is full of low-hanging fruit that only a president willing to unleash dealmakers can pick.

His own bill for service included millions of dollars in legal fees and 33 hours of congressional, special-counsel, and grand-jury testimony. Ivanka, he says, was the partner who held the Washington years together; after the move to Florida she chose to focus on raising their children and has also been making early-stage AI investments with a record he calls strong. Gil says that before they built BrainCo together he quietly asked around, because the press portrait had been so dark, and heard that both Kushners were decent, clear thinkers trying to do the right thing. Kushner's line to Ivanka at the time was that every opportunity in life has a cost, and that this one was small relative to the chance to have an impact. Because it was unpopular, they were not invited to the party, so they put their heads down; they treated each day as sand through an hourglass.

For a 25-year-old engineer staring at an offer from a frontier lab, he reaches for a line attributed to Einstein: once the mind expands, it never returns to its original size. Government, in his sales pitch, is a two-year business school. You leave a lucrative seat, you meet people from different worlds, you leave the place better than you found it, and you come back with a different lens on how the country actually works. He wrote a book after the first term mainly to recruit that traffic. The pattern he wants is the founders' pattern: leave the farm, serve, go back to the farm — today, leave the office, serve, go back to the office. A permanent political class, he says, is not in America's long-term interest.

The conversation sounds like a former official talking about states and sovereigns, but several of the screens are usable on Monday morning. BrainCo's customer test is the cleanest: is the problem complex enough that AI would create obvious efficiency, is the data actually in hand, and will the command chain force a try? Fail two of three and even a giant contract is a bad early customer. The country test is just as blunt. If the wave is not there, Kushner says you should not invest; effort does not rescue a headwind. Before you congratulate yourself on hustle, decide whether the industry and the geography are a tailwind or a slog.

Entering a domain you do not know, he used a method that looks naive and is not: ask who has already solved this, then ask who failed and why. Expert opinion is admissible, but only with the win-loss record attached. On AI inside a company, he would not start with the flashiest demo. Start where the process is long, the judgment is opaque, and the files are already digital. Construction permitting is the template sitting in plain sight: more than forty steps compressed toward three minutes, after which the value argues for itself.

The line he kept giving Ivanka is the one he leaves for people still hesitating at the edge of a public or private leap. Every opportunity has a cost. Price what you are willing to pay, then do not keep turning around to re-litigate the invoice.