loading...
doxx.net gives AI agents their own private network layer: P2P encrypted, API-controlled, and isolated from the infrastructure that tracks everything.
ENTRY ANGLES
enterprise compliance offering for regulated industries · audit-readiness infrastructure for AI agent deployments · agent identity management for SaaS platforms deploying agents · privacy-as-a-service for enterprise AI workflows
VERTICALS
CAPABILITIES
network infrastructure engineering, security and compliance expertise, enterprise sales, API development, regulatory knowledge (SOC2, HIPAA, financial compliance)
Every AI agent that runs a task on your behalf creates a trail. It calls APIs, browses pages, exchanges tokens, and leaves metadata in systems you don't control. The infrastructure the internet runs on was built for human users at human speeds; agents run hundreds of transactions per hour, using credentials issued to people and then passed down, logging nothing that distinguishes their activity from the user's.
doxx.net calls its answer Agentic Defined Networking: a private network layer that gives users and their AI agents encrypted point-to-point communication channels, configurable identity controls, and API-managed firewall and routing rules. The company provides the IP space and DNS root; agents configure the network programmatically. Its platform reports 42 million threats blocked and 13.5 million AI training samples generated.
Founded by Barrett Lyon – who previously co-founded Prolexic, one of the first distributed DDoS mitigation services, acquired by Akamai in 2014 – the company raised $38M in a Series A led by Andreessen Horowitz in October 2026, coinciding with an open beta launch.
The network model the internet runs on was never designed for agents. IP addresses resolve to users or devices, not to autonomous software processes running on behalf of users. When an agent manages your email, it does so using your credentials from your IP with your browser fingerprint. There is no native concept of agent identity at the network layer – no way to say "this activity was my agent, not me."
As AI agents take on more sensitive tasks – managing finances, processing client information, executing legal workflows – this ambiguity becomes a liability for everyone in the chain. A financial services firm that deploys AI agents to process client data needs to demonstrate to regulators that agent activity is logged, auditable, and isolated. The current infrastructure doesn't support that cleanly.
doxx.net is making a specific structural bet: that agent identity and network-level isolation will become prerequisites for deploying capable AI agents in regulated or sensitive contexts. The Andreessen Horowitz backing suggests at least one major investor agrees that the infrastructure gap is real and that this moment is early.
The most defensible near-term market isn't consumer privacy. It's enterprise compliance. The compliance frameworks that govern how financial services, healthcare, and legal firms handle data were written for human users and are being applied, imperfectly, to agent workflows. The gap between what regulators require and what current infrastructure supports creates both risk and procurement budget.
A well-positioned offering for a compliance team looks specific: agent activity runs on isolated IP ranges, is logged to immutable infrastructure, and can be audited against the user's explicit authorization scope. That's not a privacy pitch – it's an audit-readiness pitch. The buyer is a CISO or a chief compliance officer, not an individual user.
The specific entry angle worth pursuing: any enterprise currently deploying AI agents on contracted tasks – insurance claims processing, legal document review, financial compliance screening – represents an immediate procurement opportunity. These companies have both the compliance requirement and the explicit risk exposure to justify a network infrastructure purchase before the problem becomes a regulatory audit finding. The proof-of-concept pitch doesn't require a platform-wide deployment: one workflow, one isolated agent network, one audit trail the compliance team can show a regulator.