An inline authorization platform for AI agent actions
An inline authorization platform for AI agent actions
Project Details
Updated 07/31/26 · Edited by orgWe are building the inline layer that stands between AI agents and the real systems they act on. Every tool call or API action an agent attempts passes through AgentShield (the open-source gateway), gets a real-time verdict from Tenure (the decision engine) allow, block, or gate and is recorded as tamper-evident proof under OAD, the Open Agent Dossier standard.
Theory of Impact
Updated 07/31/26 · By grantmaking.aiThis Solution sits in the AI control layer, not alignment the bet that even imperfectly-aligned agents can be constrained in what they do. Most catastrophe paths route through an agent taking a real action: escalating privilege, acquiring resources, doing something irreversible. It gates those inline checks and leaves tamper-evident, signed receipts a compromised agent can't rewrite making misbehavior detectable and actions reversible-by-default.
People
Updated 07/04/26 · Edited by orgTeam Member
Funding Details
- Jul 3, 2026
- Jul 31, 2026
- -
- -
- -
- -
- -
- -
- seeking first grant
- -
working on something similar :)