Over the next few years, AI agents will handle payments, write production code, and make decisions across the internet with zero humans involved. If the trust layer underneath that expansion does not exist before it scales, the fallout will be catastrophic.
Over 40% of AI-generated code contains security vulnerabilities, and the tools producing that code are seeing hundreds of millions of dollars in annual recurring revenue from a new generation of builders who have never written a line of code by hand. It’s a recipe for disaster but there’s one startup, zauth, that is designed for addressing this massive threat.
A $500 Education in Trust
zauth’s founding story has an unconventional start. It begins with $500 in a college dorm room in 2021. The company’s founder, Dylan, got into crypto trading as a student and the space taught him hard lessons firsthand. We’re talking rug pulls, pump-and-dumps, and the real-time collapse of FTX.
“I was scammed by people I met over the internet more times than I can count,” he says. “What all of that did was teach me early that trust is the whole game. It rewired how I think about the internet and the world in general.”
That initial learning that the internet is full of systems not built to protect the people using them became the catalyst for zauth, which is an entire ecosystem designed to solve multiple trust problems at once.
Broken Pipes in the Agentic Economy
zauth did not begin as the company it is today. It started with a narrower focus: testing x402 endpoints, the infrastructure that allows AI agents to make autonomous payments across the internet. The team discovered that over 30% of those endpoints were broken. Agents were attempting transactions, hitting dead ends, losing money, and moving on without flagging any of it.
The team built a database to help agents transact safely. But the deeper they went, the bigger the problem became.
“An AI agent’s primary goal is to complete the task. That’s it,” Dylan explains. “It doesn’t pause to evaluate risk. It doesn’t flag a suspicious endpoint. It doesn’t care how it gets there. That’s not a bug that gets patched. It’s just what AI is.”
That realization transformed zauth into a full trust and security layer for autonomous AI systems.
A $20 Pentest That Outperforms the Industry
The company’s flagship product, Vector, is an AI-powered security scanner that simulates real attacker behavior to find vulnerabilities in web applications. In an industry where traditional penetration tests run into the thousands and require dedicated security teams, Vector does the job for $20. Platforms like Lovable, Replit, and Cursor have made it possible for a single person to build and ship a production web app in an afternoon. The security industry, still structured around enterprise procurement cycles and large budgets, has not adapted to that reality.
“Nobody is pentesting their Lovable app. Nobody is auditing their Cursor-generated code,” Dylan says. “They don’t know how, and until us, they couldn’t afford to.”
zauth has tested Vector against the same targets as firms that have raised tens of millions of dollars in funding and found more critical vulnerabilities. The gap between how fast people are shipping and how seriously they are thinking about security, Dylan argues, is enormous. Vector is built to close it.
Distribution Over Destination
zauth’s other major product, RepoScan, helps users make fast, informed decisions when evaluating code repositories which is a critical need in crypto trading and project research. Through integrations with platforms like Axiom, GMGN, and Terminal, RepoScan reached over 90,000 users almost overnight.
That early traction taught the team a lesson that now shapes everything they build. “It’s a lot easier to partner with a platform that already has thousands of engaged users than to try to pull those users directly to us,” Dylan says. “We’d rather be embedded in the tools people use every day than be another product they have to remember to open.”
It is the same playbook zauth is running with Vector. The goal is not to be a standalone destination but to live inside the platforms where developers already work.
The Stakes Beyond a Single Breach
Without verification infrastructure the consequences compound quickly. Agents operating at scale will not just waste money on broken services but, more critically, they will interact with compromised systems and expose private data at a pace no human attacker could match.
“AI will run the internet within the next couple of years, and the attack surface becomes the entire internet,” he says. “If we don’t build safe rails for that expansion now, while the foundation is still being laid, we end up with a world that’s ruined by the same technology that was supposed to improve it.”
zauth’s north star is straightforward but ambitious – to become a standard for trust in AI. If an agent is making a payment, touching an API, evaluating a repository, or building an application, zauth wants to be in that stack, providing accessible tooling that promotes the safe growth of the agentic internet.
The company recently won funding through the Pump.Fun Build in Public Hackathon and is putting it toward distribution, integrations, and marketing. The vision for what zauth becomes is a product that is invisible, one that lives inside the tools developers already use, tests automatically, and ships applications secure.
“We’re young, driven, and we will win,” Dylan says. “We want to work with anyone who sees what we see.”
Spencer Hulse is the Editorial Director at Grit Daily. He is responsible for overseeing other editors and writers, day-to-day operations, and covering breaking news.
