Faramesh Labs

Most agents in production have way more permissions than the job actually needs. They can call any tool, hit any endpoint, move money. You usually find out something went sideways when something goes sideways. Faramesh sits between your agent and the tools it calls, gating every action through a policy first. Permit, deny, or defer to a human. Credentials, identity, per-tool rules, audit trails, all defined as code in one file. Policies are written in FPL, our DSL with proper handling for sessi[1][2]

Visit website Suggest an update

2025 Founded1 Person on the graph

Overview

Faramesh Labs is a computer and network security company [1][3] founded in 2025 and based in New York [1]. The company develops a policy enforcement layer for AI agents [2]. According to the company's description, production agents typically retain excessive permissions, including the ability to call any tool, access any endpoint, or move money [2]. Faramesh positions itself between agents and their connected tools, evaluating each action against defined policies before execution [2]. The system can permit actions, deny them, or escalate decisions to human review [2]. Policies are configured through code in a single file, incorporating credentials, identity rules, per-tool restrictions, and audit trails [2]. The company uses FPL, a domain-specific language, for policy definition [2]. As of available records, Faramesh Labs has two employees [3].

Growth & traction

LinkedIn lists 2 people working at Faramesh Labs. [4]

Key people

Amjad F.Founder

Alumni

Amjad F.Founder

Something wrong or missing? Send an update. Fixed within 24 hours.