There is a version of venture capital that is driven by algorithm. An investor sees someone flip their LinkedIn profile to “stealth mode,” and within hours, an automated outreach tool has offered a term sheet to the top 25% of them. Mercedes Bent has watched this play out and decided to build something that looks almost nothing like it.
“We have become an index play […] and I’m basically doing the anti of that.”
That tension between venture as a relationship craft and venture as an asset-allocation game sits at the center of everything Mercedes is building at Premise, the early-stage firm she co-founded with Vanessa Larco. And the way she is building it is less about a contrarian thesis than a studied conviction that the fundamentals of the job got buried under a decade of hype.
Meet Mercedes, Meet Premise
Mercedes came to venture through a longer and more varied path than most. She worked at the Federal Reserve and Goldman Sachs during the 2008 financial crisis. She founded her own startup. She helped scale other people’s companies from early single-digit teams to organizations of 100 or more. She spent six years as a partner at Lightspeed Venture Partners, where she focused on early-stage investing and ran a scout fund starting in 2020. She also organized some of the first Ruby on Rails meetups in New York back in 2013. All of that shaped what Premise became.
“The last 15 years experience working in tech and in finance and venture capital,” she said. “I feel like it’s all accumulated and driven me to this point where I’m trying to create the VC firm of the future and really the VC firm I wish I had had access to back when I was a startup operator and founder.”
Premise focuses on pre-seed and seed investing, with a particular emphasis on founders in the AI ecosystem. The firm is small by design and deliberate by strategy.
The AI Cycle, Seen Clearly
Mercedes is not a short seller. She said so plainly and more than once. But she is also clear-eyed about what happens when a market gets ahead of itself, and she thinks the current AI moment is doing exactly that.
“The time to push in all of your chips on the table is when you see a huge new unlock for a technology start to happen,” she said. “We’re in a hypey moment of AI, but in the grand scheme of things, this is such early innings. More akin to when mobile was coming out than it is to the end of the mobile cycle, which was like 2021.”
Her concern is not that AI is overstated as a long-term technology. It is that the inflated valuations and accelerated capital deployment of this moment are creating a distorted picture of what these businesses actually look like. She pointed to the potential IPOs of OpenAI and Anthropic as the likely moment of correction. Once lockup periods expire, typically six months after an IPO, public market investors start applying fundamentals-based scrutiny that private market investors have not been required to apply.
“Public market investors are not VCs,” she said. “And they’re just gonna keep looking and saying, so this thing loses how much money for every dollar of revenue it makes.”
She outlined several other potential triggers beyond an IPO correction: large-scale enterprise security incidents, political backlash tied to job displacement, and shifts in public sentiment if unemployment spikes in ways that become visible and politicized. She noted that the U.S. has a presidential election in 2028 and that job loss related to AI automation could become a defining campaign issue if the effects reach beyond the tech sector.
Her read on what early-stage investors should do in this environment is simple: resist the pull to underwrite the exception.
And Drew adds: “A lot of investors have been underwriting the anomaly instead of underwriting the median,” citing cases where investors price companies based on the assumption they will reach $100 million in revenue in seven months.
The discipline is to return to realistic unit economics and to stop letting hype set your expectations.
Building a Deal Flow Engine That Does Not Look Like One
At the center of Premise’s sourcing strategy is Surreal, a community Mercedes started as a monthly dinner series for engineers and AI researchers. The first gathering happened in her living room. It has since moved to art galleries, where attendees sit among large paintings and sculptures and talk about the latest developments in AI research, agent deployment, and model efficiency.
The community now has over 5,200 applicants in the past year, with hundreds of regular attendees. Most of them, Mercedes said, do not know there is a VC behind it. That is the point.
“The minute you’re saying, hey, here’s office hours for VCs, come and talk about your idea, now people are posturing,” she said. “I wanna get to know the real you, who is real-time working out in their head as they’re sitting there, a software engineer at Vercel thinking I might go start a startup one day.”
She thinks about the community through the lens of product design, specifically the sequence of “magic moments” that keep people engaged. The first happens when someone shows up at a dinner, sits down next to a person with remarkably similar interests, and spends two hours in conversation that produces a potential hire, a potential co-founder, and a possible customer. The second hits when they join the Surreal WhatsApp group and discover that 10 people they already know are in it. The third arrives when they post a problem at 11:59 p.m. and someone answers it immediately.
“I want to draw out the magic experience for you to just keep coming back and feeling like it’s amazing,” she said.
The tool running much of the back-end logistics is Luma, which Mercedes described as far better suited for business-oriented community management than the alternatives.
The Behavioral Economics of Good Judgment
Mercedes applies behavioral economics as both a diligence lens and a personal operating discipline. On the diligence side, she looks for founders who understand the emotional architecture of their product.
“The founders who are most in tune with behavioral economics actually build the most magical products.”
If a founder can explain what the user will feel when they sign up, how that cascades through the experience, and how it translates to key performance indicators, that is a signal worth paying attention to.
On the personal side, she uses it to manage her own decision-making. The biggest risk she flags is getting emotionally captured by one part of a deal, usually the founder, and letting that excitement override judgment about everything else.
“I get so excited about founders sometimes on the very first meeting with them and I think it through and I’m like, well, it’s a terrible deal,” she said. “Maybe they’re just good at talking.”
The training is to make sure the founder, the product, the market opportunity, and the deal terms all earn conviction independently. When one of them is missing, the discipline has to override the enthusiasm.
She also flagged the opposite failure mode: letting a deal that worked despite its flaws become a template for future decisions. Chasing an outlier as though it represents a repeatable pattern is, she noted, its own form of cognitive error.
The Long Game, Played Up Close
Mercedes is building Premise for a version of the market that requires patience, presence, and a willingness to know founders long before they are ready to take a meeting with a VC. Her bet is that the firms built on relationship density rather than deal volume will survive cycles better, and will earn the trust of the founders worth backing.
“I wanna see that whole journey with you,” she said. “And have you naturally come to it.”
In a market where capital flows fast and algorithms optimize for reach, that kind of long-horizon, in-person relationship work is genuinely rare. Whether the AI bubble deflates through an IPO correction, a political shift, or something else entirely, Mercedes is positioning Premise to be one of the firms still standing when the next real cycle begins.
This season is supported by SVB. Silicon Valley Bank, a division of First Citizens Bank. Member FDIC. SVB is a trusted collaborator for the founders pushing boundaries and the investors who back them. We’re proud to have them as our sponsor.
Please note, this podcast is for informational purposes and is not investment, financial, or legal advice. The views expressed are those of the speakers and do not necessarily reflect the position of SVB.
More from Uncovered Media
