Artificial intelligence is reshaping not only how networks operate but also how networking companies are built. A new startup called Doxx is betting that AI agents can design, deploy, secure and manage private networks with far less human intervention than traditional infrastructure requires. The company is launching an Agentic Defined Networking platform and has raised a $38 million Series A round led by Andreessen Horowitz, with participation from Animo Ventures and Focal.vc.

The launch marks a new chapter for founder Barrett Lyon, a serial entrepreneur who has spent more than two decades building and selling networking and security companies. His previous ventures include Prolexic, defense.net and Netography, which were acquired by Akamai, F5 Networks and Vectra AI, respectively. Doxx argues that the same AI wave that is changing applications and operations can also change the economics of building a global network.

Key facts at a glance

• Doxx is launching an Agentic Defined Networking platform that lets users and AI agents build private networks.
• The company has raised $38 million in Series A funding led by Andreessen Horowitz, with Animo Ventures and Focal.vc participating.
• The platform combines custom VPN transports, an encrypted mesh and a separate control plane.
• Supported transports include WireGuard and custom options over WebSocket, SIP, ping, QUIC, a public CDN layer and HTTPS.
• Doxx says the network is run entirely by AI across 31 locations worldwide.
• AI agents receive identities, certificates and scoped tokens so they can perform network tasks without holding master account credentials.

A serial networking founder turns to agentic infrastructure

Lyon's career has been defined by building infrastructure companies that solve hard network problems. Prolexic focused on distributed denial-of-service protection before cloud-based mitigation became standard. Defense.net continued work in network security, and Netography explored network detection and response. Each company was eventually acquired by a larger infrastructure or security vendor. Doxx is different in both ambition and architecture: it is not just a security control or a monitoring layer, but an attempt to let AI operate the underlying network itself.

The founder said the project began as a hobby. He created a TCP tunnel that could run over a public content delivery network and help traffic avoid detection and censorship. That experiment, known as Darkflare, drew enough interest to push him toward a larger idea: a VPN that cannot easily be stopped and that is autonomously managed by AI.

Lyon described the progression simply: the hobby became an obsession, and the obsession became a business. The result is a platform that tries to combine the resilience of overlay networking with the automation of AI agents. Instead of requiring a large engineering team to configure every tunnel, route and certificate, Doxx aims to let agents handle much of the operational work under human oversight.

How the Agentic Defined Network platform works

Doxx's platform is built around three main pieces: custom VPN transports, a company-run mesh and a separate control plane. The design is meant to give users private connectivity that can traverse restrictive networks while remaining manageable through APIs and AI agents.

Covert transports

The client supports WireGuard as well as custom transports for VPN traffic over WebSocket, Session Initiation Protocol, ping, QUIC, a public CDN layer and HTTPS. These transports are designed to get through firewalls and other network filters that might block conventional VPN protocols. By disguising or adapting traffic to common protocols, Doxx aims to keep connectivity available in environments where censorship or strict filtering is common.

Mesh and addressing

The mesh connects every site to every site with encryption. It runs over IPv4 and IPv6 and is orchestrated through an application programming interface. 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 directly, without the address translation barriers that often complicate direct connections.

Control plane, DNS, certificates and BGP

Lyon said the client app is a device, not a control plane. Routing, port forwarding and firewall rules are managed in a separate portal. That separation matters because it keeps the endpoint client from becoming an all-powerful management console. The company also runs its own Domain Name System root with 196 domains and its own certificate authority. Lyon said Doxx wrote its own Border Gateway Protocol daemon and runs internal BGP across the mesh. Those choices give the platform control over naming, trust and route propagation, which are usually spread across multiple external services.

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. The design is intended to reduce the amount of personal information stored in a central directory while still making it possible for devices to discover and authenticate one another.

How AI runs the network

The need for AI to run the network is pragmatic. Lyon said he turned to AI because he could not manage the network by himself. The platform has 31 locations around the world, and internally there is a mesh network. As a human being, he said, he cannot manage all of it alone.

Before building, the team modeled every site down to the wire and the optic in a virtual model. 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. The process shows how AI can participate not only in day-to-day operations but also in the physical buildout of infrastructure.

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 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 a 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 said users should not simply give agents their master account credentials. The scoped-token model is intended to limit what an agent can do if it is compromised or misconfigured.

The name is a reversal, not a threat

The term doxxing usually refers to publishing someone's private information, but that is the opposite of what Doxx is trying to do. Lyon registered the name more than 30 years ago because it was his hacker name at the time. He said he is reclaiming the word, and he finds it funny that the company is doing the exact opposite of doxing. For him, the name is fitting.

The contrast is also a useful reminder of how networking culture has changed. Early internet communities often used handles and pseudonyms, and privacy was enforced through technical knowledge and trust networks. Modern infrastructure, by contrast, depends on certificates, identity systems and policy controls. Doxx's address book, certificate authority and agent tokens reflect that shift: the company is trying to preserve privacy while building enterprise-grade controls around identity and access.

Enterprise plans and open source

Looking forward, the plan is to build out more enterprise offerings and release some code as open source. The company believes the technology is not just about using AI for networking but also about using AI to build the technology itself. That distinction matters because it suggests a feedback loop: AI agents help operate the network, and the network in turn provides the environment for agents to build and run services.

Lyon said entrepreneurs can have a vision of where a technology will go, but that vision often hits a wall when it reaches personal limitations. AI, he said, takes the work well beyond that point. Running the company with an agentic partner, in his view, turns the effort into something much more powerful. The next phase for Doxx will test whether that model can scale from a launch platform into durable enterprise infrastructure.

Source:Network World News