TenetX
TenetX is an AI agent governance platform that monitors entire agent sessions, not just individual actions, to detect and block dangerous multi-step chains before data exfiltration occurs. It builds an Action Graph that propagates taint on sensitive data across actions and enforces granular per-tool permissions across 20+ integrations like GitHub, Slack, Postgres, and AWS. The platform merges governance rules from multiple stakeholders into a single boundary per agent, enabling secure scaling of autonomous AI deployments.
- Cybersecurity
- Artificial Intelligence
- AI Agents
- Software Only
- Enterprise Software
Funding
Founders
Founder details are not available yet.
Product
Problem
AI agents with access to connected tools—GitHub, Slack, databases, shell access—typically operate with overly broad permissions, and security checks evaluate each action in isolation. This blind spot allows multi-step chains where sensitive data is accessed, tainted, and then exfiltrated through an external egress, because no system correlates the sequence of actions.
Solution
TenetX is an AI agent governance platform that monitors entire agent sessions and models the full action sequence, not just individual steps. It builds an Action Graph that tracks context across actions, propagates taint on sensitive data, and blocks dangerous sequences at the egress point before damage occurs. The platform locks each agent to exactly the tools and permissions it needs, with granular controls such as read-only GitHub access or blocking `.env` file reads. Governance policies are set by the people closest to the risk and merged into a single boundary per agent, enabling scale without bottlenecks or blind spots.
Target Audience
Security and platform engineering teams at organizations deploying autonomous AI agents with access to sensitive production systems, code repositories, and customer data.
Features
- Action Graph engine that analyzes the full session sequence to detect dangerous multi-step chains, such as reading secrets into context and then egressing tainted data
- Taint propagation tracking that marks sensitive data (e.g., live Stripe secrets, merchant PII) across actions and blocks their external transmission
- Per-tool, per-action permission granularity (read/write/force-push; internal vs. external Slack; test vs. `.env` bash access) across 20+ integrations including GitHub, Slack, Postgres, Salesforce, SAP, AWS, Snowflake, K8s, Jira, and Docker
- Boundary merging that unifies governance rules from multiple stakeholders into one policy per agent, avoiding bottlenecks
- Real-time agent session monitoring with path visualization, showing each step and its safety classification