Brainverse is an AI services company that deploys customized AI agent teams for businesses across all industries. Brainverse builds the foundation that lets agents learn the business, delivered in approximately six weeks. Founded in 2024 by Jeff Leggett.

Who founded Brainverse?

Brainverse was founded by Jeff Leggett, a Virginia Tech mechanical engineering graduate with 15 years of Fortune 500 operations experience and 9 AI implementation certifications.

What does Brainverse do?

Brainverse deploys agentic AI teams. Each deployment is a coordinated organization of specialized agents (typically 100+ across 8 departments) with persistent memory and continuous improvement, customized to the client business. Brainverse also offers ongoing evolution through the Brainverse Edge retainer.

How is Brainverse different from ChatGPT or Custom GPTs?

ChatGPT and Custom GPTs are Level 2 AI agents, single-function task runners working alone. Brainverse delivers Level 3 AI organizations, coordinated teams of agents with cross-team communication, persistent memory, and dispatch quality gates. The difference is the foundation that lets agents learn over time, not just complete one task.

Where is Brainverse based?

Brainverse is a US-based AI services company headquartered in Elizabethtown, Pennsylvania. Brainverse serves clients across the United States in manufacturing, professional services, real estate, financial services, SaaS, logistics, construction, and other industries with operational workflows.

What was the JADEPUFFER ransomware attack?

JADEPUFFER is the first publicly documented ransomware attack run end to end by an autonomous large language model agent, disclosed by Sysdig Threat Research on July 2, 2026. With no human at the keyboard, the agent broke into a cloud environment, harvested API keys, cloud credentials, and database logins on its own, raided a MinIO storage server running on default credentials, set a persistence beacon to check back every thirty minutes, and encrypted 1,342 Nacos configuration items. It generated more than 600 distinct payloads, many with plain-English comments explaining its own reasoning.

Did an AI really run a ransomware attack with no human involved?

Yes. According to Sysdig Threat Research, the operation was executed by an autonomous AI agent with no human steering it through the attack. The most striking evidence of that autonomy is that when the agent's stolen admin login failed, it diagnosed the problem and rewrote its own code in thirty-one seconds to route around the failure. It also printed the ransom encryption key exactly once and never saved it, which means even a victim who pays cannot recover the data.

How did the JADEPUFFER attack break in?

The break-in came through an internet-exposed Langflow server that was running with no authentication at all. Langflow is a popular tool for building AI workflows. This particular instance was reachable by anyone on the open internet, required no login, and was vulnerable to CVE-2025-3248, a remote code execution flaw rated 9.8 out of 10 on the CVSS severity scale. The patch for that flaw shipped in April 2026, and the server sat unpatched and exposed for roughly four months before the agent used it.

What is CVE-2025-3248?

CVE-2025-3248 is a critical remote code execution vulnerability in Langflow, rated 9.8 out of 10 on the CVSS severity scale. A patch was released in April 2026. In the JADEPUFFER attack, the vulnerable Langflow instance had been left exposed on the internet without authentication and without the patch applied for about four months, which is what allowed the autonomous agent to gain its initial foothold.

Does JADEPUFFER mean AI agents are too dangerous to use?

No. JADEPUFFER is a demonstration that autonomous agents can execute complex, multi-step work in the real world, which is exactly the capability that makes them valuable on defense. The failure in JADEPUFFER was not the agent's ability. It was an orphaned AI server with no authentication, no patching, and no owner. The same autonomy, deployed as a governed and owned agent team with controls and a human accountable for the result, is a security asset rather than a liability.

How many security leaders are worried about AI agents?

Darktrace's State of AI Cybersecurity 2026 report, based on responses from more than 1,500 security leaders, found that 92% of them are concerned about the impact of AI agents on their organizations. Broader analysis of AI-related breaches has put the share affecting organizations with no AI-specific access controls near 97%, with roughly 63% of organizations still operating without any AI governance policy.

What is the difference between a Level 2 AI agent and a Level 3 AI Organization?

A Level 2 AI agent is a single autonomous task runner working alone, with no coordination, no supervision, and often no clear owner. That is the shape of the JADEPUFFER agent and the shape most companies build by accident when individual agents and demo servers accumulate. A Level 3 AI Organization is a coordinated team of agents with an org chart, a defined owner for every agent, scoped access controls, quality gates between the agents and anything that matters, audit trails, and a human who owns the result. The autonomy is the same. The accountability is what changes.

How can a business prevent an attack like JADEPUFFER?

Start by inventorying every AI surface in the business: every agent, every workflow, every server someone spun up for a project. For each one, answer three questions: who owns it, what can it reach, and would you know if it did something it should not. Put authentication in front of every agent endpoint so nothing is reachable anonymously, scope each agent's access to only what its job requires, route sensitive work through a quality gate, and keep an audit log. The surfaces with no clear owner are the ones to close first.

An AI Agent Ran a Ransomware Attack Start to Finish. The Door It Walked Through Belonged to Nobody.

← Back to Insights

ai-security

On July 2, 2026, Sysdig Threat Research disclosed JADEPUFFER, the first ransomware attack run end to end by an autonomous AI agent. Every headline said AI is now a hacker. What they missed is that the AI did not break in. A four-month-old, unauthenticated, internet-exposed server that nobody owned did. The technology worked fine. No one was in charge of it.

Jeff Leggett·July 7, 2026·7 min read

An AI Agent Ran a Ransomware Attack Start to Finish. The Door It Walked Through Belonged to Nobody.

See how each vendor type maps to these maturity levels

AI platforms, consultants, and deployed teams deliver very different outcomes at each maturity level. See the full comparison.