Headline and Key Facts

• Headline: Startup Doxx.net hands the network controls to AI.
• Launch: Doxx.net launched this week with a platform for Agentic Defined Networking, allowing users and their AI agents to build private networks.
• Funding: $38 million Series A led by Andreessen Horowitz, with Animo Ventures and Focal.vc participating.
• Founder: Barrett Lyon, who previously founded and sold Prolexic to Akamai, Defense.net to F5 Networks, and Netography to Vectra AI.
• Origin: The project grew out of Darkflare, a hobby TCP tunnel over Cloudflare designed to evade detection and censorship.
• Architecture: Custom VPN transports, a company-run mesh, and a separate control plane.
• AI operations: The entire network, spanning 31 locations worldwide, is run by AI because Lyon said he could not manage it alone.
• Agent tools: An agent gateway gives AI agents identities, certificates, and tokens; agents can run BGP, create domains, sign certificates, and host applications.
• Identity: The address book contains no phone numbers, email addresses, or names; users exchange cryptographic keys and assign aliases.
• Infrastructure: Doxx.net runs its own DNS root with 196 domains, its own certificate authority, and its own BGP daemon.
• Name: Lyon registered the domain more than 30 years ago as his hacker name; the company is reclaiming the term doxxing by doing the opposite.
• Future: Plans include more enterprise offerings and releasing some code as open source.

AI Takes the Wheel in Networking

It used to take a large team of engineers years to build a new networking startup. AI changes that game entirely. That is the premise behind Doxx.net, a new venture that launched this week with a platform for what it calls Agentic Defined Networking. The platform lets users and their AI agents build private networks. The company also announced a $38 million Series A round led by Andreessen Horowitz with participation from Animo Ventures and Focal.vc.

The startup is the latest project from Barrett Lyon, a serial entrepreneur who has founded and sold multiple networking vendors over the last two decades. His previous companies include Prolexic, which was acquired by Akamai; Defense.net, acquired by F5 Networks; and Netography, acquired by Vectra AI. Now he has turned his attention to AI with a platform that promises to hand the network controls to autonomous agents.

A Serial Founder's Path to Agentic Networking

Lyon's career has been defined by building infrastructure that defends, routes, and observes network traffic. Prolexic became a major name in DDoS mitigation before Akamai acquired it. Defense.net focused on cloud-based DDoS protection and was picked up by F5 Networks. Netography brought a cloud-native approach to network security and visibility, and Vectra AI acquired it to deepen its AI-driven threat detection portfolio. Those exits gave Lyon a rare view into how networks are built, attacked, and defended at scale.

Doxx.net has its roots in the Darkflare project, which Lyon created as a hobby. Darkflare was a TCP tunnel over Cloudflare that could tunnel through public networks in a way that helps avoid detection and potential censorship. The response to Darkflare led Lyon to consider a VPN that cannot be stopped and is autonomously managed by AI. The hobby turned into an obsession, and the obsession turned into a business, Lyon said.

What Agentic Defined Networking Means

Agentic Defined Networking is a departure from traditional software-defined networking. In SDN, a centralized controller manages network behavior through programmable interfaces. In an agentic model, AI agents become active participants. They can provision connectivity, request addresses, manage routing, issue certificates, and deploy services. The network is still controlled by policies and tokens, but the day-to-day operation is delegated to software agents that understand intent and can act on it.

Doxx.net combines custom VPN transports, a company-run mesh, and a separate control plane. The client application is not the control plane. Routing, port forwarding, and firewall rules are managed in a separate portal. That separation is important because it keeps the device focused on connectivity while administrative functions remain centralized and auditable.

Inside the Doxx.net Architecture

Covert Transports

The Doxx.net client supports WireGuard and custom transports for VPN traffic over WebSocket, Session Initiation Protocol, ping, QUIC, Cloudflare, and HTTPS. These transports are designed to get through firewalls. The goal is resilience: if one protocol is blocked, another can carry traffic. That approach echoes the Darkflare origin, where tunneling through public networks was a way to avoid detection and censorship.

Mesh and Control Plane

The mesh connects every site to every site with encryption. It runs over IPv4 and IPv6 and is orchestrated through an application programming interface. Because the mesh is fully connected, any site can reach any other site without relying on a central hub. The control plane remains separate, allowing administrators to manage routing, port forwarding, and firewall rules from a portal rather than from the endpoint client.

Addressing, DNS, and Certificates

Devices on the mesh are not behind network address translation or carrier-grade NAT, according to Lyon. That allows peer-to-peer applications to work more naturally. The company runs its own Domain Name System root with 196 domains and its own certificate authority. Lyon said the team wrote its own Border Gateway Protocol daemon and runs internal BGP across the mesh. Running its own DNS root, CA, and BGP daemon gives Doxx.net control over naming, trust, and routing inside its network.

Identity and Discovery

The address book holds no phone numbers, email addresses, or names. Users exchange cryptographic keys and assign an alias to each contact. A directory service seeds a new device with the data needed to find its contacts. After that initial call, devices sync with each other. This design puts identity in keys rather than in personal identifiers, which reduces the amount of personal data the network needs to store.

How AI Runs the Network

The need for Doxx.net to be run by AI is pragmatic. Lyon said he turned to AI because he could not manage the network by himself. The whole thing is run completely by AI, he said. Behind the scenes, it is a network with 31 locations around the world, and internally there is a mesh network. As a human being, he could not manage all of it alone.

Lyon said the team modeled every site, down to the wire and the optic, in a virtual model before building. An infrastructure management system running the company's own AI on its own hardware then ordered the installation. It orchestrated shipping and delivery through data center APIs. Human technicians performed the remote smart hands installations. That combination of virtual modeling, AI-driven procurement, and human physical labor shows how AI can extend a small team's reach across global infrastructure.

Agents as Network Operators

The company also built tools for agents to work with users and with the network. An agent gateway gives an AI agent an identity in the Doxx.net chat app. The user pastes a credential into the agent. The agent obtains its certificate and appears in the user's chat. Users can create group chats with several agents. In one example, Lyon said one agent runs BGP while others handle other tasks.

The API is self-teaching. An agent given the configuration endpoint can learn how to build networks without further instructions. It then asks what the user wants to do, such as link an office to a house or give a Linux box a private IP address. Agents can also create their own domains and host names, sign certificates, and run web servers and applications inside the network.

Administrators issue read-only or admin tokens for each agent. Tokens can expire or be revoked. An administrator can also cut an agent's gateway access or delete the agent in the app. Lyon emphasized that users do not just give agents their account credentials to the master account. That token model is a guardrail. It allows agents to act with limited authority, and it gives administrators a way to revoke access if something goes wrong.

From Darkflare to Doxx.net

The term doxxing usually refers to publishing someone's private information, but that is the exact opposite of what Doxx.net is about. Lyon registered the domain more than 30 years ago, as it was his hacker name at the time. He said he is reclaiming the word, and it is kind of funny that the company is doing the exact opposite of what doxing is. It just seems fitting for him.

That inversion is more than a naming story. The platform's identity model is built to minimize exposure of personal data. The address book has no phone numbers, email addresses, or names. Users rely on cryptographic keys and aliases. The network uses its own DNS root and certificate authority. Those choices reflect a philosophy of controlling the infrastructure that establishes trust and discovery, rather than depending on external identifiers.

Enterprise Plans and Open Source

Looking forward, the plan is to build out more enterprise offerings as well as release some code as open source. Fundamentally, the technology is not just about using AI for networking but also about using AI to build the technology itself. Lyon said a person can have a vision of where something is going to go, and then that vision hits a wall when it meets personal limitations. AI takes it well beyond, running the effort with an agentic partner, so to speak, and turns it into fire.

The $38 million Series A, led by Andreessen Horowitz with participation from Animo Ventures and Focal.vc, gives Doxx.net room to expand. The company's bet is that AI agents will not just assist network engineers; they will become operators with identities, certificates, and scoped permissions. If that model works, the next generation of private networks may be built and run by a small human team working alongside a much larger fleet of AI agents.

Source: Network World News