David Minarsch didn’t want to build another centralized AI platform. He watched the same infrastructure patterns repeat themselves across tech’s landscape: first came the dominance of a few massive companies, then came the creators who built on those platforms only to discover they’d built on someone else’s land.
Today, the Olas founder is solving that problem at the intersection of cryptocurrency and artificial intelligence. Through Pearl, his consumer-facing AI agent application, Minarsch is enabling everyday users to build, own, and profit from their own autonomous agents.
What started as academic research into multi-agent systems has evolved into a movement toward what Minarsch calls “true ownership” in the digital world.
The connection to Roblox (NYSE:RBLX) provides the clearest window into Minarsch’s philosophy. Just as the game development platform created millionaires from teenage coders without formal technical training, Pearl aims to democratize AI agent creation for anyone with an idea.
Don't Miss:
- Accredited Investors Can Now Tap Into the $36 Trillion Home Equity Market — Without Buying a Single Property
- From Moxy Hotels to $12B in Real Estate — The Firm Behind NYC's Trendiest Properties Is Letting Individual Investors In.
"When people realize that AI doesn't have to belong to Big Tech, but can belong to them, that's when ‘decentralized ownership’ stops sounding abstract and starts feeling personal," Minarsch told Benzinga.
David Minarsch Draws Inspiration From Roblox To Build A User-Owned AI Ecosystem
Minarsch points to Roblox as proof that open access can turn ordinary users into builders. The platform's success reinforced his conviction that accessibility and ownership can redefine how technology ecosystems grow.
The contrast between Roblox and what Olas is building comes down to control. Roblox owns the network, while Olas distributes it.
“Our goal with Pearl is similar and more ambitious, still,” Minarsch told Benzinga. “Like Roblox, we're focused on making creation accessible to everyone — but instead of virtual worlds, it's the world of AI agents.”
That goal guided every design decision behind Pearl, Olas' consumer-facing app. The product keeps the simplicity of Web2 while removing the friction that often deters newcomers from cryptocurrency-based tools. On Pearl, users can skip wallets, bridging, and token swaps entirely.
“Users can simply use their debit/credit cards to fund their agents, sign in through familiar social logins, and use natural language to instruct their agent with their goals using their own words, rather than crypto jargon,” Minarsch said.
The approach has already produced tangible results, Minarsch told Benzinga. Early beta participants have built personalized portfolio-management agents by customizing general blueprints through conversation.
To expand that momentum, Olas introduced a $1 million Accelerator program funded in USD Coin, designed to support emerging "supercreators" developing new agent templates across finance, entertainment, and lifestyle.
Unexpected Uses For AI Ownership On Olas
Few things surprise Minarsch anymore when it comes to what builders create when freed from centralized constraints. Two projects supported through the Olas Accelerator capture just how broad that creative range can be, Minarsch said.
The first, called Cupid Agent, blends matchmaking with coaching. Built by graduate students at Imperial College London, it learns about users through natural conversation and their digital footprint, then suggests compatible matches through AI-driven dialogue.
The second project, PettAI, explores gaming instead of relationships. The companion agent manages digital pets in virtual worlds, automating repetitive actions like feeding and resting so players can focus on gameplay.
"Both the Cupid and PettAI use cases target entirely different audiences but share a core principle: solving real human needs through user-owned AI," Minarsch told Benzinga.
The Olas Accelerator supports up to 10 development teams with grants reaching $100,000 each. The selection process focuses on identifying builders capable of identifying real user problems and solving them effectively.
Minarsch said he prioritizes clarity above other factors. Successful candidates articulate precisely what problem their app addresses, identify the specific users needing solutions, and outline distribution strategies for reaching those audiences.
The evaluation also considers realistic development timelines, Minarsch told Benzinga. “We also care about the product distribution these teams already have achieved and the total addressable market size of their product,” he said. “Although especially with respect to the latter, we're fairly lenient on as many autonomous AI agent products are green-field innovations where sizing the market is often a guessing game, and it's equally important to see what excitement the product can create once live."
The Centralization Problem That Millions Of Users Face But Don’t Discuss
Ownership in the digital space has become rare, Minarsch said. "We live in an era where we don't really own anything anymore. From music and movies to software and even digital identities, everything's become subscription-based or tethered to someone else's platform. For younger generations, especially, ownership feels like a distant dream and even a foreign concept," he added.
Minarsch finds that idea troubling, especially as AI becomes central to daily life. He often uses real-world examples to make the concept concrete. "Sometimes that means explaining it like real estate — why owning your home is better than renting your home," he said. "Other times, it's using pop culture analogies, like how Taylor Swift taught an entire generation about the value of owning your own IP."
Developers often describe building “on” OpenAI or other centralized platforms. Technically, Minarsch argues, they’re renting. When an infrastructure provider identifies a market opportunity that your app has proven successful, you become competition rather than a customer.
“Nobody truly ‘builds on OpenAI.' They rent its APIs. And when you rely fully on a centralized platform, you risk being replaced by it,” Minarsch told Benzinga. “Builders want foundations that won't undermine them, and that's what a decentralized ecosystem like Olas offers. It's open, composable, and community-owned.”
Trending: An EA Co-Founder Shapes This VC Backed Marketplace—Now You Can Invest in Gaming's Next Big Platform
Agent-To-Agent Transactions Reach 8.6 Million Payments Using Coinbase x402 And Ethereum Standards
Among the least discussed yet most consequential innovations within Olas exists in the infrastructure enabling machines to compensate other machines for services. This capability, termed agent-to-agent transactions, has already executed over 8.6 million such exchanges, according to Minarsch.
Supported by technologies such as Coinbase's (NASDAQ:COIN) x402 payment standard and Ethereum's EIP-3009 proposal, agents can handle their own payments for data, storage, computation, or content without human approval.
The economic model supporting this activity fundamentally reimagines how digital services function. Traditional architecture requires subscription bundles for access to application programming interfaces, even when actual usage remains sporadic. This bundling model works acceptably when humans make occasional requests, but it becomes economically irrational when autonomous systems require microsecond access to diverse services, Minarsch said.
“A prediction market trader agent might need access to multiple data sources and computing resources,” Minarsch told Benzinga. “Rather than each user obtaining separate API credentials from each provider, agents pay for services as needed through micropayments.”
This architectural change reduces friction substantially, he said. Users fund their agents through straightforward mechanisms, and then the agents autonomously purchase necessary services through small cryptocurrency transactions. Data providers receive immediate payment for their services, and the entire transaction occurs without human intermediation.
Olas' Chat Mode And Natural Language Remove Crypto Barriers For Mainstream AI Agent Users
Pearl’s beta version deliberately targeted cryptocurrency-native users, but the industry's complexity remains a barrier for mainstream adoption, Minarsch said.
“A lot of people who are crypto-curious never actually use crypto because it’s too complex,” he told Benzinga. “Bridging, on-ramping, wallets, [decentralized exchanges], and a dozen unfamiliar terms make it intimidating. Pearl removes all that friction.”
That's why Olas built features like Chat Mode and Co-Pilot Mode that address this issue by allowing users to describe their intentions in simple language. The agent then does the work behind the scenes.
"Look at AI in general, it has been around for years, but only took off when packaged in the simple, familiar text-based ChatGPT interface," Minarsch said. "The goal of AI is to solve a real problem with value-added simplicity in a familiar layout."
With 413 daily active agents on Pearl, obvious questions emerge about whether the platform has achieved genuine product-market fit or simply early adoption from a niche community. Minarsch measures this through a specific metric: weekly retention of active agents.
“The goal is to make Pearl’s [user experience] so intuitive and seamless that it becomes an everyday part of users’ lives, no different from Apple’s App Store,” Minarsch said. “We see agents as the natural evolution from apps, and when we see that sustained shift in usage from apps to agents, that’s when we’ll know we’ve found genuine product-market fit.”
There’s also a network effect to consider. As Pearl user growth accelerates, new specialist AI agents emerge in the Olas Mech Marketplace that existing Pearl users can employ. According to Minarsch, more specialist agents increase utility for existing users without anyone needing to coordinate the expansion, an organic growth pattern that indicates ecosystem health.
What Prevents Olas From Becoming Another Centralized Platform Like OpenAI? Full-Stack Code Ownership, Minarsch Says
Critics often raise the same concern about decentralized systems: without clear safeguards, any platform can drift toward the same centralization it was created to avoid.
Most platforms offering consumer ownership actually concentrate control at the developer level, with hosting and infrastructure management remaining centralized, Minarsch said. Users might control wallet addresses without controlling the actual system code.
Olas takes a different approach, Minarsch told Benzinga. Pearl users receive complete access to the underlying code running their agents, which represents full-stack ownership rather than partial control.
Minarsch described it as a living cycle where every new specialist entering the marketplace expands the network's reach and hands more power into the hands of its users.
“When local capabilities aren’t enough, or when a model or compute exceeds what’s locally available, the marketplace comes into play,” Minarsch told Benzinga. “Our agents rely on other specialist agents in the background to execute tasks. As the user’s agents become better through access to more specialist agents via the marketplace, these specialists compete against other specialists attracted by a flourishing demand-side, and the whole Olas ecosystem becomes stronger and more decentralized."
When asked what discussion he wishes filled more media coverage, Minarsch’s answer was straightforward: ownership.
“Ownership and why it’s so important. It’s absurd that it doesn’t get discussed more,” he said. “We’re seeing some conversation about countries building ‘sovereign AI,’ but that doesn’t represent a sufficient solution because the infrastructure will still depend on a few centralized entities.”
See Also: The ‘ChatGPT of Marketing' Just Opened a $0.81/Share Round — 10,000+ Investors Are Already In
Decentralized AI Infrastructure Must Be Built Now Before Big Tech Achieves Complete Dominance
Minarsch started researching autonomous AI agents in 2019, before most technology leaders took them seriously. Then, between 2019 and 2023, very few people understood what agents could become. He was drawn to them because he recognized their potential as economic participants enabled by blockchain’s new financial rails.
The calculus changed dramatically, Minarsch said, adding that the AI economy is growing exponentially on dangerously centralized infrastructure. If a more democratic, decentralized model is desired, the time to build it is now, while the industry remains relatively young.
“Someone needs to be a trailblazer, right?” Minarsch told Benzinga. “The time is now, while we’re still in relatively early days. Not five years later, when all the infrastructure is built, and we’ve merely perpetuated Big Tech's dominance and control in the Web3 era."
The Three Principles That Define Pearl’s Autonomous Agent Platform
Minarsch framed Pearl as a product built on three guiding principles: ownership, curation, and transparency. Each user retains full custody of their agents and personal configurations, a structure meant to ensure lasting control rather than temporary access.
The platform's curated library gathers agents designed to handle real-world needs across many areas of life. Through an open-source, independently audited codebase, Pearl exposes every layer of how those agents function, a model Minarsch described as the groundwork for technology people can truly call their own.
"This road doesn't end here, though," Minarsch said. "Our long-term vision is to bring you the most powerful AI agents that you can truly own, be it on mobile, desktop, or custom device. We're just getting started."
Read Next: This Jeff Bezos-backed startup will allow you to become a landlord in just 10 minutes, with minimum investments as low as $100.
Image: David Minarsch
