Autogon is a platform focused on AI security protection, and its products are built around one core idea: the cost of detection is decreasing, while verifiable evidence is what matters. The platform currently includes four products, each addressing security issues at a different layer.
Nemesis Shield is a runtime protection product for applications and servers. It is integrated through a one-line SDK and supports Python, Node, Go, Ruby, PHP, Java, .NET, Rust, as well as browser-side (React/Angular/Vue/jQuery) and Supabase Edge. The SDK computes a privacy-preserving form of the request locally (method, normalized route, whether the caller is authenticated) and does not upload request bodies, keys, or source code. Shield first learns the real behavioral baseline of each tenant application in observation mode, then switches to enforcement mode after user confirmation, blocking requests that deviate from the baseline in-process. It also monitors networks and cloud environments, correlating attacks against applications, DNS, and the cloud into a single event. Shield uses a positive security model and can catch zero-day vulnerabilities, broken object level authorization (BOLA), and business logic abuse, and protects unpatched legacy frameworks. The browser-side SDK can address the client-side controls required by PCI DSS 4.0.1. Shield supports self-service use, offers a free tier, can protect an application in about two minutes, and observation and enforcement modes can be switched in the console without redeployment.
Autogon Omniguard is a financial crime detection engine for banks, fintechs, and payment companies. It scores every transaction in real time, screens counterparties against global sanctions, PEP, and adverse media data, verifies identities, and generates suspicious transaction reports in the goAML format used by financial intelligence units worldwide. The product can adapt to different markets' currencies, identity systems, and law enforcement lists, and is built according to Nigeria's CBN baseline standards, and can also be configured for other regulatory requirements.
Nemesis Blue is a hardened endpoint security agent that individual users can use for free. It runs through the security channels provided by each operating system (EndpointSecurity on macOS, eBPF on Linux, ETW on Windows), avoiding system crashes caused by kernel driver updates. Blue identifies attacks based on behavior rather than only known signatures, and immediately cuts off an infected device's connection to the attacker after flagging it. Its releases are Ed25519-signed and hash-verified, and it has self-check mechanisms.
Nemesis Red is an autonomously operating penetration testing engine. It can autonomously map systems, discover weaknesses, actually exploit vulnerabilities to confirm their authenticity, and even discover entirely new vulnerabilities. Red emphasizes that every finding must be verified on a third-party scoreboard before it counts, in order to avoid false positives. In controlled tests, the number of confirmed issues was nearly ten times that of an ordinary one-off scanner (p<0.001), and it once discovered and reported a new memory safety vulnerability in an open-source library.
The four products can be used independently or combined to cover the full security chain from application runtime, money flows, and endpoints to vulnerability discovery. Blue and Red have been officially released, and Shield and Omniguard support self-service use and start for free. The platform provides a trust center that publicly discloses complete architecture specifications, threat models, and data egress details.