Your NHI strategy has a
blind spot
AI agents are the fastest-growing class of non-human identity in your organization. They access databases, call APIs, make decisions, and interact with other agents, but they're invisible to your current NHI tools.
Traditional NHI platforms weren't built for this
The NHI market is focused on managing service accounts, API keys, OAuth tokens, and SSH keys. Platforms like Oasis, Entro, Astrix, and Clutch do this well, but AI agents represent a fundamentally different class of non-human identity.
AI agents don't just authenticate and execute a fixed operation. They reason, make decisions, call tools dynamically, and interact with other agents. They connect to MCP servers whose tool surfaces can change without notice. They operate with capabilities that drift over time.
Traditional NHI platforms can discover that an agent has an API key. But they can't answer the questions that matter: What capabilities does this agent actually use at runtime? Has its behavior drifted from its declared purpose? Which MCP servers is it connected to, and have their tools changed? If this agent is compromised, what's the blast radius?
Purpose-built NHI governance for the agentic era
AIM isn't a bolt-on. It's designed from the ground up for AI agent identity governance.
Agent-native identity
Traditional NHI tools manage service accounts with static tokens. AIM provides cryptographic identity purpose-built for AI agents.
- Ed25519 keypairs with challenge-response auth
- Post-quantum cryptography ready (ML-DSA)
- Agent framework auto-detection
- MCP server attestation with drift detection
Open source
The AIM repository is public under Apache-2.0, and it can be self-hosted with Docker Compose.
- Apache-2.0 licensed, full source on GitHub
- Self-host or use AIM Cloud
- No vendor lock-in, no per-seat minimums
Governance by default
AIM doesn't just discover agent identities. It governs them from creation to retirement.
- Ownership attribution to human owners
- Capability-based access control
- 9-factor trust scoring algorithm
- Audit log of action checks
AIM vs. enterprise NHI platforms
See how AIM compares to traditional NHI platforms for AI agent identity governance.
| Capability | Traditional NHI | AIM |
|---|---|---|
| Open Source | ||
| Pricing | Enterprise (contact sales) | Open source |
| AI Agent Identity | Bolt-on | |
| Cryptographic Agent Auth | ||
| MCP Server Attestation | ||
| Capability-Based Access | Generic RBAC | |
| Trust Scoring | Basic risk | 9-factor algorithm |
| ABOM Export | ||
| Self-Hosted Option | ||
| Service Account Management | Complements |
Note: AIM is designed to complement traditional NHI platforms, not replace them. Oasis, Entro, Astrix, and Clutch manage service accounts, API keys, and OAuth tokens across cloud environments. AIM governs the new class of NHI that those platforms weren't designed for: autonomous AI agents. Many enterprises will use both.
Complete NHI lifecycle for AI agents
From discovery to retirement, AIM manages every stage of the agent identity lifecycle.
Discover
Find shadow agents across your environment
Register
One-line SDK with auto-detection
Govern
Capability-based access control
Monitor
Trust scoring and drift detection
Retire
Automated lifecycle transitions
One line for developers. Full visibility for security.
Developer simplicity meets enterprise governance.
Developer
from aim_sdk import secure
agent = secure("my-agent",
capabilities=["db:read", "api:call"]
)
@agent.perform_action(capability="db:read")
def get_customer(id):
return db.query(id)Security Team Sees
Governance reports
Reports for agent NHI governance reviews.
NHI inventory report
Complete census of all agent identities, owners, capabilities, and lifecycle status
ABOM export
CycloneDX-compliant export of agents, MCP connections, and dependencies
Access audit trail
Best-effort row per action check that passes its signature check: agent, capability, resource, stored trust score and whether it was allowed
Ownership attestation
Prove every agent has an accountable human owner with review dates
Start governing agent NHIs
Choose your deployment model and get started immediately.
AIM Cloud
# Sign up at aim.opena2a.orgpip install aim-sdkFree plan with limits on hosted agents, MCP servers and team members. No infrastructure required.
Try AIM CloudSelf-hosted
git clone https://github.com/opena2a-org/agent-identity-management.gitdocker compose up -dFull control on your infrastructure. Docker or Kubernetes.
View on GitHubStart with a scan
npx hackmyagent secureScan your agents for vulnerabilities first.
Try HackMyAgentFrequently asked questions
A non-human identity is any digital entity that authenticates and operates without direct human control: service accounts, API keys, OAuth tokens, bots, and AI agents. AI agents are the newest and fastest-growing category of NHI. Unlike traditional NHIs that execute fixed operations, AI agents reason, make decisions, and call tools dynamically, requiring a different approach to governance.
Traditional NHI platforms like Oasis, Entro, and Astrix focus on managing service accounts, API keys, and OAuth tokens. They're excellent at what they do. AIM is purpose-built for AI agent identities specifically, with cryptographic authentication, MCP server attestation, capability-based access control, and agent-native trust scoring. AIM is open source under Apache-2.0.
AIM complements existing NHI platforms rather than replacing them. Traditional platforms manage service accounts and API keys. AIM manages AI agent identities. Many enterprises will use both: their existing NHI platform for traditional machine identities, and AIM for the agent identity layer.
AIM is open source under the Apache-2.0 license, including self-hosting. AIM Cloud has a free plan with limits on hosted agents, MCP servers, team members and requests a month; a self-hosted AIM server has no plan limit. Single sign-on with Google, Microsoft Entra ID and Okta is planned. Self-hosted AIM can forward audit log entries to Splunk HEC. This forwarder is not available in AIM Cloud.
AIM's Python SDK ships integrations for LangChain, CrewAI and MCP servers. An agent built on another Python framework, AutoGen among them, registers and wraps its actions through the same SDK, and registration records the framework it finds in the agent's imports as the agent type; there is no AutoGen-specific integration. AIM also has a Java SDK, and a TypeScript SDK is published as @opena2a/aim-sdk.
An ABOM is a machine-readable inventory of an AI agent's identity, capabilities, MCP server connections, and dependencies, similar to an SBOM for software. AIM generates ABOMs in CycloneDX format for compliance and supply chain transparency.
The open-source NHI platform for agentic AI
Enterprise NHI platforms are proprietary, priced for large deployments, and treat AI agents as an afterthought. AIM is purpose-built and open source.