About
Making AI systems safe and compliant, and giving you the evidence to prove it.
Our Mission
AI systems powered by large language models can fail in ways traditional software never could. These failures occur inside the agentic system where decisions are made, tools are invoked, and outputs reach users.
Superagent's mission is to make AI systems safe and compliant, and give teams the evidence to prove it. We provide defense models that prevent failures at runtime, continuous tests that measure safety, and a status page that proves it to customers and prospects.
The Team
Alan Zabihi (CEO) and Ismail Pelaseyed (CTO) are the co-founders of Superagent.
Before language models existed, they built a startup that generated Google text ads based on e-commerce inventory using proprietary algorithms. Customers included some of the largest e-commerce companies in the Nordics, such as ICA, the dominant groceries retailer. The company was acquired.
They have experience from companies including Google, SKF, Klarna, and Trustly. This work provided exposure to how enterprises handle security, procurement, and deployment processes for software systems.
They have also built several popular open source projects in the AI space, including one of the first open source AI agent frameworks, which was adopted by enterprises building agentic experiences. They also built VibeKit, a sandboxing and security layer for coding agents, and the unofficial coding agent CLI for xAI's Grok models.
Backed by the Best
Superagent is backed by Y Combinator, Rebel Fund, and the founders of Replit, Okta, and HubSpot.