A new platform called ADIN uses AI agents to autonomously analyze and invest in startups, completing tasks like due diligence and valuation in about an hour instead of weeks. Its creators believe AI can bring a data-driven "moneyball" approach to venture capital, improving success rates and lowering costs by moving beyond human intuition.
The main topics covered are the emergence of AI-driven venture capital platforms, the debate over whether venture investing is a quantifiable science or an intangible art, and the potential for AI to disrupt the traditional VC industry.
Prominent VC Marc Andreessen represents the opposing view, arguing that venture capital is an art form based on intangible factors and guiding founders, making it resistant to full automation. This highlights a central tension within the industry about AI's future role.
Last fall, as venture capitalists were sinking record sums into artificial intelligence, a group of investors gathered to appraise a new startup. The company, Infinity Artificial Intelligence Institute, made software to automatically tune AI models, making them faster and cheaper. The founding team seemed strong, and the market was rapidly expanding. Half of the investors were cautious; the other half saw dollar signs. One of them dubbed the deal an “absolute banger.”
This startup was real, and so was the $100,000 the VCs invested in its seed round. But the VCs themselves were all AI agents, part of a new platform called ADIN, the Autonomous Deal Investing Network.
Launched in 2025, ADIN uses AI to replace the human analysts involved in venture dealmaking. Put in a startup’s pitch deck, and out comes a detailed analysis of its business model and founding team, a list of diligence questions and compliance risks, an estimate of the total addressable market, and a suggested valuation. ADIN has about a dozen different agentic investors, each with a distinct persona and investing thesis. The Tech Oracle looks at a startup’s underlying technology; the Unit Master evaluates the financial fundamentals; the Monopoly Maker, loosely based on Peter Thiel, looks for market dominance. When the majority of the agents like a startup, they suggest how much ADIN’s fund should allocate to the deal. The platform does this in about an hour, compared to the days or weeks that it takes an analyst at a VC firm.
“The game of venture doesn’t have a high success rate,” says Aaron Wright, the cofounder of ADIN’s parent company Tribute Labs. The current approach—a kind of finger-in-the-air, gut intuition about who and what will become the great unicorns of tomorrow—yields “home runs,” where a startup returns 10x or more of the invested capital, only about 1 percent of the time. Three-fourths of venture deals don’t even recover the cost of capital.
As Wright sees it, AI models could significantly improve those odds. He believes venture capital is entering its moneyball era, where quantitative methods overtake human intuition, and everyone starts to hit more home runs. “Increasingly, these systems will be able to eliminate bad projects, focus on those that are more successful, and also lower the cost of operating some of these enterprises,” Wright says. In a matter of years, he believes AI agents could be some of the best venture investors in the world.
And when that happens? “There may be no more Sand Hill Road.”
Few groups of people are more bullish on AI than venture capitalists, who collectively invested more than $200 billion into the AI sector last year. Advancements in AI models have transformed the way investors think about nearly every company, in nearly every industry. Vinod Khosla, the founder of Khosla Ventures, recently predicted that AI will replace 80 percent of job responsibilities by 2030. Yet many venture capitalists seem to underestimate the extent to which AI may impact their own jobs.
Marc Andreessen—the celebrity venture capitalist and cofounder of Andreessen Horowitz—said on an episode of his podcast, The Ben & Marc Show, that when AI is doing everything else, venture capital may be “one of the last remaining fields that people are still doing.” The job is more than just writing checks, he argued; it’s also choosing the right ideas, at the right time, with the right people, and then guiding them to success.
“That’s not science, that’s art,” Andreessen continued. “If it was a science, you could eventually have somebody who just dials it in and gets 8 out of 10. But in the real world, it’s not like that. You’re in the fluke business. There’s an intangibility to it. There’s a taste aspect.”
Many of the VCs I spoke with for this story took Andreessen’s line. Keval Desai, managing director at the venture capital firm Shakti, likened early-stage investing to “picking Michael Jordan in kindergarten.” When a startup is young, there’s no product or revenue—just potential. “You can have all kinds of compute, all kinds of algorithms, but without data, there is nothing to analyze,” Desai says. (Still, he copped to occasionally prompting Gemini to “act as a VC analyst” when considering unfamiliar markets.)
Brian Nichols, the cofounder of Angel Squad, an angel investing network associated with the early-stage VC firm Hustle Fund, told me he wouldn’t trust AI to do the “curation” work of investing. At the end of the day, VC is a business of networks: It’s about who you know, and who you can personally vouch for. At the same time, he thought AI could probably replace other parts of the job. When we spoke, he had just returned from a Hustle Fund off-site where a partner had used Claude Code to build a tool to triage founder emails. “We all spend hours every day responding to founder pitches,” he says. “Those are hours that could probably be spent otherwise.” Aydin Senkut, founder and managing partner of the VC firm Felices, told me he thinks most VCs are experimenting with AI in some capacity to try to remain competitive. His firm is currently experimenting with chatbots to write investment memos, improve deal sourcing, and help partners to “score” startup founders.
Projects like ADIN seek to automate even more of this lower-level work. The diligence process—where investors investigate a startup’s viability, risks, and growth potential—is one of the most time-consuming parts of venture capital, especially when investors are considering companies in nascent markets. ADIN collapses this step into a few minutes, and can quickly surface regulatory or compliance issues that could torpedo a deal. When assessing a mining technology company, ADIN flagged a series of export control laws and cross-border data transfer issues. “These aren't the kinds of questions most investors think to ask,” says Priyanka Desai, one of ADIN’s partners. She added that the AI “doesn't get tired, doesn't have blind spots from habit, and surfaces the long tail risks that are easy to overlook.”
For now, humans are still required for a few things. ADIN’s dealflow, for one, comes from a network of venture scouts. While ADIN is funded like a traditional VC fund, with limited partners, it offers an unusual financial compensation to its scouts, who can receive 50 percent of the carried interest, or profits, that are usually reserved for general partners. “It's basically like giving away general partner level economics to someone just to submit a deal and leverage their network,” says Desai.
Humans are also required for the “last mile,” including meeting founders and ultimately deciding whether or not to cut them checks. “We know these systems are not perfect, so you want to have double checking,” says Wright. The AI agents can sometimes be overzealous in their recommendations: He showed me a startup that the agentic VCs had all loved, but where ADIN decided not to invest after meeting with the founders and running into concerns with existing competitors.
On the other hand, Wright says, he’s also used ADIN to evaluate startups that already have $20 million or more in funding, some of which ADIN’s agents uniformly dislike. “Our challenge is to understand, is that accurate or is that an error?” he says. In some cases, it’s possible investors have fallen into a common human trap: hyping up a startup or a founder on vibes alone.
Whether or not AI systems outpace investors is one thing. But another existential threat remains: The same AI technology that could make venture work faster and more efficient is also making it easier and cheaper to start software companies. The VC industry has made the majority of its money in the last decade from software as a service. But a startup that once needed a $2 million seed round to hire a specialized engineering team might now achieve the same product velocity for less than six figures with a handful of vibe coders. The math no longer favors the big checks.
Until recently, only a tiny fraction of unicorns were bootstrapped. The average software unicorn raised $370 million, according to SaaStr, which monitors SaaS companies. Now, there are startups like the AI image generator Midjourney, which reached unicorn status with a core group of employees. (According to the latest data from Pitchbook, Midjourney has around 100 employees. Court documents related to a copyright lawsuit peg the company’s annual revenues at more than $300 million. Midjourney did not respond to WIRED’s request for comment on figures obtained from Pitchbook.)
This particular scenario—where some founders no longer need venture money at all—is the one likeliest to scare venture capitalists. “That’s the existential threat,” Nichols, the Angel Squad founder, says. “The money’s there, but the need from the founder isn't.” Perhaps AI won’t outright replace investors. But it could make those investments unnecessary.
Aside from companies in robotics, biotech, or other hardware-focused industries, fewer startups may soon need the kind of mega funding that the venture capital industry has been built on. That could return the industry to its origins: a small, specialized field that bridges the gap between scientific breakthroughs and their commercial applications. (The giant companies building foundational models would fit in here, and would likely continue to take VC money to pay for the eye-watering amounts of money they spend on compute, data centers, and employee compensation.)
If it’s possible to build a startup for cheap, we could see the industry become leaner—fast. That could put investors out of a job in a different way: not by replacing them, but by replacing their business model. “If you have these funds sitting on their hands, competing over the very few deals that actually need funding, that creates a different type of problem,” Nichols says. “This is actually what investors are losing sleep about.”
Meridiano
