Lineation
When you let an AI agent act on its own — call APIs, touch databases, send emails — you lose the ability to watch over its shoulder the way you would a human employee. Lineation sits between your agents and the systems they touch, checking every action ag
🔗 Visit Lineation
Description
Giving an AI agent the keys to your systems is a bit like hiring a very fast new employee who never sleeps and never asks permission — useful, but unnerving if you can't see what they're doing. Lineation is built to fix that: it puts a checkpoint between every AI agent and the tools/data it's allowed to touch, so a mistake or a hijacked prompt gets stopped before it turns into a real incident.
Lineation is a security control plane for autonomous AI agents, giving each agent a distinct "non-human identity" (NHI) under zero-trust rules rather than letting agents share generic API keys. It inspects prompts and tool calls in real time to catch prompt injection, redacts PII and secrets before they leak, and runs agent actions inside a sandboxed environment. It ships an MCP gateway and multi-provider LLM gateway (OpenAI, Claude, Google, Azure), policy-as-code governance, RBAC, and immutable audit trails with replay — with compliance templates aimed at SOC 2, HIPAA and the EU AI Act. Pricing is usage-based on protected agent actions, from a free 5,000-action tier up to a $1,399/month Team plan and custom Enterprise (including on-premises deployment).
💬 Our review
The short version: if your company is moving from AI chatbots to AI agents that actually take actions — calling APIs, touching customer data, executing workflows — Lineation is a guardrail layer built specifically for that risk, not a generic security add-on retrofitted for agents.
Against general-purpose LLM gateways like Portkey or OpenRouter, Lineation's edge is that it treats each agent as its own identity with its own audit trail, rather than just routing traffic — that's the difference between logging what happened and actually being able to stop a bad action before it executes. The free tier (5,000 actions/month) is generous enough to evaluate seriously, but the jump to $249 then $1,399/month means costs can scale fast if your agents are chatty. Worth adopting once you have agents touching production data or regulated workflows; probably unnecessary for a single internal prototype bot with no real access.
💰 Pricing
📊 Global score
🤖 AI-enriched data
Gratuit jusqu'à 5 000 actions protégées/mois. Growth 249 $/mois (50k actions). Team 1 399 $/mois (1M actions). Enterprise sur devis (déploiement on-premises inclus).
Pros
Identité non-humaine dédiée par agent (zero-trust) plutôt que des clés API partagées
Détection en temps réel des prompt injections et redaction PII/secrets
Passerelle MCP + multi-LLM avec audit immuable et replay
Modèles de conformité prêts pour SOC 2, HIPAA, EU AI Act
Palier gratuit généreux pour tester (5 000 actions/mois)
Cons
Tarification à l'action peut devenir coûteuse à grande échelle
Catégorie encore jeune, pas de recul long terme sur la fiabilité
Pas de date de fondation publique — difficile d'évaluer la maturité de l'équipe