From fintech to foundational models, ex-Square and Nearside founder Thomson Nguyen wants to shape AI’s “innocent Wild West” with operator-first VC.
I’d never heard someone frame their personality as as “orthogonal” to others’, let alone frame orthogonality as the unique selling point that makes a quarter-billion-dollar venture firm’s leadership structure work in lockstep, but, lo, that all changed on October 9th during a video call with Thomson Nguyen, Co-Founder and Managing Partner of Saga Ventures. Nguyen, alongside Max Altman (of Alt Capital and family fame) and Ben Braverman (Flexport’s CRO emeritus), manages a $125 million early-stage venture fund, adding operator expertise and engineering know-how to Saga’s work.
Nguyen comes to Saga by way of fintech: Following a leadership role on the Data Science/ML team at Square Capital, and an entrepreneurial residency at Kleiner Perkins, Nguyen founded a micro-SMB-focused neobank, Nearside (f.k.a. “Hatch”), in 2019. Plastiq acquired Nearside in 2022 for $130 million. Having raised around $55 million from firms like Kleiner Perkins, Valar Ventures, and Foundation Capital, Nguyen said the exit was an “okay outcome.”
“It wasn’t the big swing that I wanted,” Nguyen said. In hindsight, he realized that the highly regulated nature of fintech made it hard for Nearside to take off — enforcement actions against partner banks and other headwinds “added external delay” that had “nothing to do” with Nearside’s fundamentals.
During a brief post-exit hiatus in New York, Nguyen, who was interviewing for GP roles at larger firms, began talking with Braverman and Altman, friends of his, about a joint move into venture investing; Altman was just finishing up at Alt Capital, and had been thinking about starting his own office. They noticed their complementary skillsets — Nguyen on the technical side, Braverman in sales, Altman on the institutional capital end of things — and felt their combined strengths would bode well for early-stage investment.
The less regulated nature of AI, alongside the “orthogonal” personalities and skillsets of the Saga Ventures co-founders, helped shape much of the firm’s initial thesis. “We help startups graduate from Seed to Series A in the best path possible: through getting your first engineers in the door, visualizing your sales motion, your go-to-market, and how you sell your startup, and then when it comes time to raise your Series A, we introduce you to all the tier-one investors,” he said. Portcos under the Saga partners’ belt run the gamut: fairly staid investments on one end, like Ramp, and, perhaps for pacifists, more controversial prospects on the other, like Anduril.
Nguyen seems to advocate for something akin to the “move fast break things” ethos of the late aughts. Framing the current state of AI development as an “innocent Wild West,” he said substantial regulation and safety layers — the kinds that were eventually implemented to regulate the internet writ large — are probably five to 10 years away, meaning the current moment is a “foundational period where we just need to see what’s possible first.”
“I still think there’s so much room to run with these models, these companies, and these products before we can start to understand the shape of how they interface with daily human life,” he said. It sounds a bit like opening Pandora’s box to take a peek at what’s inside; Nguyen said he sees LLMs already capable of replicating entire classes of jobs.
But opening Pandora’s box is unprecedentedly expensive. The spider web of all-stock deals and tit-for-tat arrangements among OpenAI and outfits like Nvidia, Microsoft, and others eerily echo the creative accounting behind the dot-com bubble — to the tune of $1 trillion. Landmark litigation like Anthropic’s $1.5 billion settlement for alleged copyright infringement further suggests top-of-funnel fundamentals — as in, the costs associated with legally gathering data to build models — aren’t getting any cheaper.
The myriad costs involved in developing AI systems may explain the value behind one of Saga’s more successful portcos: Profound, which helps brands get LLMs to mention them more often. (Think search engine optimization, but for Claude, not Google.) The existence of startups like Profound, alongside recent product launches like ChatGPT’s Pulse, suggest a new revenue layer is forming, the same way ads have defined and funded private-sector internet activity for decades. For better or worse.
“There is a scenario in which we try to close Pandora’s box and say, Hey, maybe the internet was a bad idea, guys, let’s just all go home. We can’t do that for the internet. We can’t do that for LLMs and AI either,” Nguyen said. “In every technological breakthrough in the last century and a half, what we’ve always seen is that, while you can’t close Pandora’s box, you have to actually then reset to a net-new normal.”
Despite that quasi-resignation to apparently “inevitable” forces — a framing many tech theorists contest, if not outright deride — Nguyen calls himself an optimist. For instance, some may see the advent of generative video platforms like Sora as a troubling orbital jump that will further erode public trust in the visuals we encounter online; Nguyen does not support “slop for slop’s sake,” but thinks that, for every 4000 to 5000 videos, a user could be inspired to become the next James Cameron (of Avatar). He also thinks a backdrop of low-quality AI videos can make people appreciate high-quality, human-generated cultural contributions — like something by Puccini; Nguyen’s an opera fan — more than they already do. And, taking a retrospective stance, he thinks technology has served several industries well, including the field of journalism, where individuals can earn a living by striking out on their own and becoming newsy influencers. (Which I disagree with, looking at the growth of local news deserts, media layoffs, and mistrust of facts.)
Whatever the ultimate outcome of this multi-trillion-dollar investment in AI — civilizational collapse? a dot-com-y redux? cybernetic nirvana? — Nguyen said he remains focused on practical applications of technology that can help address workaday pain points. He’s interested in making it easier for people to start local businesses, looking at the ways AI can accelerate the permitting process for restaurant owners and other mom-and-pop ventures. “If you look at the history of developed countries and how they’re able to build from zero to where they are… it is almost certainly through a combination of the right social planning and public policy, which we can’t control as venture capitalists or technology operators,” Nguyen said as a way to explain the potential for technology to shape how policies affect people in real life.
This interest in policy, livelihood, and technology may mean Nguyen’s portcos get even harder to define over time. “At this point, it’d be really hard to bucket any company or any person into a theme: All companies are fintech companies, but also all companies are AI companies now too,” he said.
