01 · Identity (IAM)
Verifies Who
Establishes calling identity at login or token issuance. Necessary, but static.
Category Framework · Executive Briefing
The Runtime Authorization Gap
Identity verifies who. Authorization verifies what.
Runtime Authorization verifies every action.
The Non-Human Execution Gap
Critical Enterprise Blind Spot
Entra ID / Okta Verified
↓
Non-Deterministic Agent
⚠ Everything below happens WITHOUT human identity context
MCP Tool Invocations
Function payloads & schema calls
REST & GraphQL API Calls
Outbound system modifications
Database Writes & Deletes
Direct SQL queries & table mutations
Financial Transactions
Transfers, refunds, purchase orders
Agent-to-Agent Delegations
Sub-agent task dispatching
Third-Party SaaS Actions
Salesforce, SAP, ServiceNow edits
The evolution of identity and access security
From static access to intelligent, runtime authorization.
Era 01
IAM
Human user identities & SSO directories.
Authenticate users
Establish identity
Era 02
PAM
Privileged credentials & vault access.
Protect privileged access
Secure credentials
Era 03
API Security
Rate limiting, routing & token inspection.
Secure APIs
Control traffic
Era 04
AI Gateways
Prompt filtering & text guardrails.
Filter content
Reduce AI risks
Era 05
Runtime Authorization
Action-by-action parameter & authority enforcement.
Enforce authority in real time
Every action, every time
The Unanswered Question
Enterprises have invested millions in security tools.
✓ IAM
✓ PAM
✓ API Security
✓ SIEM
✓ Zero Trust
Yet none of them can answer this single question:
“Should this autonomous agent perform this specific tool call, with these parameter values, under whose authority, right now?”
Category Requirements
What a Runtime Authorization platform must do
A credible runtime control layer must do more than inspect prompts or validate tokens. It must understand authority, evaluate actions in context, enforce decisions inline, and preserve evidence.
01
Understand Business Authority
Know the approved purpose, owner, monetary limits, permitted resources, and operating conditions of the AI agent.
02
Inspect Every Invocation
Evaluate tool calls, API requests, agent-to-agent delegations, and database actions before execution.
03
Evaluate Parameters
Assess payload values, transaction amounts, destinations, schemas, and contextual risk rather than relying on coarse scopes.
04
Enforce Inline
Permit, restrict, redact, step up, limit, or deny the action directly in the execution path without breaking agent workflows.
05
Produce Verifiable Evidence
Bind the authority, policy, parameters, decision, and outcome into tamper-evident records for audit and investigation.
Only then is runtime authorization complete.
These requirements define the control layer enterprises need between autonomous reasoning and real-world execution.
The IntentGate Category Solution
The Complete Runtime Authorization Platform
IntentGate fulfills the Runtime Authorization paradigm through two commercial products and one built-in evidence capability.
01 · Control Plane
IntentGrant
Governs business authority, monetary limits, and ownership before runtime.
02 · Data Plane Engine
IntentGate
Enforces tool calls and parameter schema limits at sub-millisecond speeds.
Included with IntentGate
Built-in Proof™
Binds decisions to cryptographically signed (Ed25519) non-repudiable logs.
Establish Runtime Authorization for your AI workers.
Book an architecture briefing to see how IntentGate fills the non-human execution gap in your enterprise estate.