Luke Day

In 2017, twelve repeat founders pooled their own capital - $1.5 million total - and made a quiet bet that venture capital had the incentives backwards. No external LPs. No institutional backing. Just a dozen people who had built things, failed at things, and raised money from people who hadn't done either. They called it MAGIC Fund.

Luke Day was one of them.

He wasn't a lifelong VC. He came up through engineering recruiting - finding and assembling the technical teams at Discord and Roblox, two of the most talent-obsessed companies in consumer tech. That background gave him something unusual for an investor: a finely calibrated sense of who actually builds things and who just talks about it.

MAGIC Fund's Fund I returned 5x over three years. The fund backed Retool before it became a unicorn. It invested in Mono - later acquired by Flutterwave - and in dozens of companies across Africa, Latin America, Southeast Asia, and North America that larger VCs couldn't be bothered to visit. By July 2021, the firm raised a $30M Fund II, with check sizes of $100K-$300K and a portfolio spanning 30+ countries.

The structure is unusual by design. MAGIC Fund runs as a decentralized network of founder-GPs - twelve investors, each an expert in their sector and geography, each making autonomous decisions with the fund's capital. There's no investment committee in the traditional sense. Instead, each GP operates from deep domain knowledge: Luke Day owns gaming, metaverse, and sports.

The idea behind MAGIC Fund traces back to something Naval Ravikant said - that the best founders, the ones who have real options when choosing investors, specifically want money from people who've been in the arena. Not advisors. Not analysts who ran models in business school. People who have shipped products and faced payroll.

So the twelve co-founders did something unusual: they put their own capital in first. Every one of MAGIC Fund's twelve GPs contributed personally to Fund I. The result wasn't just a proof of concept - it was a signal. If you're asking other founders to trust you with their companies, you start by trusting the thesis yourself.

Luke Day came to this through engineering recruiting - a discipline that requires understanding both what founders actually need (great engineers) and how to identify people who can deliver it. At Discord during 2018, he helped assemble engineering teams at a company that was growing faster than almost anyone could staff. At Roblox from 2019 to 2022, he did it again at an entirely different scale - a company that had quietly become one of the most complex engineering environments in consumer tech.

That experience meant something specific as an investor: Day understood what it looks like when a founding team can execute, and what it looks like when they can't. Technical recruiting, at its best, is a form of due diligence - interviewing people under pressure, testing for depth versus polish, figuring out who ships and who narrates.

Luke Day's corner of MAGIC Fund's portfolio sits at the intersection of participation and identity - where people don't just consume products but inhabit them. Gaming is no longer a vertical; it's a distribution layer, a social graph, and increasingly a financial system. The metaverse is the contested space where those things converge. Sports is the oldest community-building technology humanity ever invented.

That framing - communities that people care about enough to pay for, argue about, and organize their lives around - shapes how Day thinks about early-stage investments. The companies he backs tend to be infrastructure plays, community platforms, or founder-led consumer products with unusually strong retention from day one.

His engineering recruiting background matters here too. Evaluating a gaming or metaverse startup at the seed stage means figuring out whether the technical team can actually build what they're describing - whether the architecture holds up, whether the founder understands the product constraints. Day has been having those conversations for years, just from the other side of the table.

MAGIC Fund's portfolio reads like a deliberate tour of the parts of the global startup economy that institutional VC tends to miss. There's Retool, the internal tooling company that became a unicorn in 2021. There's Mono, a Plaid-for-Africa play that was acquired by Flutterwave in early 2026. There are fintech companies in Southeast Asia, developer tools in Europe, and healthcare startups that operate in markets most US investors have never visited.

The geographic distribution isn't incidental - it's the product. A fund with twelve autonomous GPs, each embedded in their region, ends up with a portfolio that looks nothing like what a single-city firm would build. The common thread isn't sector or geography; it's founder quality and the conviction that experienced builders can spot other experienced builders faster than anyone else.

Most venture investors arrive with a recognizable pedigree: Stanford MBA, two years at McKinsey, stint at a larger fund, then go solo. Luke Day arrived through engineering recruiting at Discord and Roblox - two companies that were, at the time, among the most aggressive engineering hiring machines in consumer tech.

That path is actually better preparation than it sounds. Engineering recruiting at Discord in 2018 meant operating inside a company that was growing faster than almost any infrastructure team could handle - learning to assess technical talent under pressure, to think about team composition from first principles, and to understand what it actually takes to build software products at scale.

At Roblox, from 2019 through late 2022, the context shifted. Roblox had become a different kind of company - a platform with hundreds of millions of users, a complex creator economy, and one of the most technically demanding product surfaces in gaming. Understanding what engineers needed to build there gave Day a specific lens on what gaming and metaverse startups actually require to succeed.

He departed Roblox in November 2022, moving into full-time work at MAGIC Fund. By then, the fund had already proven its thesis on Fund I. The question for Day and the other GPs was whether the model could scale to $30M and 110+ companies without losing what made it work in the first place - the founder-to-founder chemistry that no institutional process can replicate.