AI Agent Runtime Governance

AIエージェントを、本気で仕事に使うために。

散らばるAIの権限をひとつに。承認なしの操作は、もう卒業。

No credit card · 1-minute setup · CASA Tier 2 assessed

An AI agent connects safely to Gmail, Calendar, Slack, Jira and HubSpot through the VESS AIdentity approval gate. AI AIdentity

Secure AI Tool Calling with AIdentity

Every high-stakes action waits for a human approval — so AI never goes rogue.

Without AIdentity

The agent sends on its own. Confidential info leaks.

you: Reply to Alice at ACME — please thank her for the proposal.
AI Agent: Pulling context from Slack, HubSpot, and internal docs...
AI Agent: Drafting email to alice@acme-corp.example.com...
--- email body ---
To: alice@acme-corp.example.com
Subject: Re: Proposal — thanks!
Hi Alice, thanks for the proposal.
FYI — we're acquiring Y Inc. on 6/3 (unannounced); this PoC factors that in.
Looking forward to next steps.
--- end of email ---
AI Agent: ✓ Email sent to alice@acme-corp.example.com
  (no approval prompt · no preview · no audit log)
 
Confidential M&A plan just left your company.
No approval. No audit trail. No way to undo.
With AIdentity

You approve before it sends. The leak is blocked.

you: Reply to Alice at ACME — please thank her for the proposal.
AI Agent: Pulling context from Slack, HubSpot, and internal docs...
AI Agent: Drafting email to alice@acme-corp.example.com...
--- email body ---
To: alice@acme-corp.example.com
Subject: Re: Proposal — thanks!
Hi Alice, thanks for the proposal.
FYI — we're acquiring Y Inc. on 6/3 (unannounced); this PoC factors that in.
Looking forward to next steps.
--- end of email ---
VESS: gmail.send requires approval
  Preview shown to you · External recipient · Contains unannounced M&A Approve (y/n)?
you: n
Blocked — email never sent.
Every AI action captured in the audit log. Nothing leaked.

How it works

AIに任せて、あなたが承認するだけ。

日々の業務はこの4ステップ。重要な操作は必ずあなたの承認を通ります。

メールで依頼

「このお客様に見積もり返信して」とAIに頼む

AIが下書き

文脈を集めて返信メールを自動作成

あなたが承認

送信前にプレビュー → 承認 / 却下

実行&記録

承認後に送信。全操作が監査ログに残る

Features

Core Features

Simplified AI Agent Identity Management

AI Integration (MCP)

Seamless connection with Claude Desktop and AI applications via Model Context Protocol

お使いのAIに1行で接続:

$ claude mcp add aidentity https://app.vesslabs.ai/mcp

Service Integration

Unified access management for external services like Google, GitHub, and Slack

Credentials

Securely manage and verify AI agent permissions and access rights

Audit Log

Complete audit trail of all AI agent activities — who did what and when

Agent Authentication

Issue unique IDs to each AI agent for secure authentication without centralized systems

Delegation Management

Flexible permission delegation across projects with approval workflows and verifiable credentials

Use Cases

Applicable to various scenarios

AI Agent Permission Management

Grant different permissions to multiple AI agents and securely control access to tools and resources.

  • Project-level access control
  • Per-tool permission settings
  • Temporary permission issuance and revocation

External Service Integration

Convert OAuth tokens to credentials for unified access management to external services.

  • Gmail, Google Drive access
  • GitHub repository operations
  • Slack message delivery

Security & Trust

Enterprise-grade security backed by independent third-party assessment.

CASA Tier 2 Certified

Certified

AIdentity has successfully passed an independent CASA (Cloud Application Security Assessment) Tier 2 review, conducted by a third-party assessor as part of the Google & CSA CASA program. CASA uses the OWASP Application Security Verification Standard (ASVS) as its baseline. This certification verifies our security posture across authentication, authorization, data protection, cryptography, and incident response.

  • Independent third-party security assessment (zero Critical / High / Medium findings)
  • Aligned with OWASP ASVS — authentication, session management, access control, cryptography
  • Verified for handling Google OAuth-restricted scopes (Gmail, Calendar) for AI agent integrations

Pricing

Simple pricing in USD. Cancel anytime.

Free

For individual testing

$0
  • 1 user / 1 project
  • Slack, Gmail, and Jira access
  • Approval required for every write action
  • Up to 100 write actions per month
  • Audit log retention: 7 days
  • Unlimited read actions

Team

For teams operating AI across real workflows

$49 / seat / month (3 seats minimum)
  • Team-wide policies
  • Approval routing + role management
  • 10,000 write actions per month (team total)
  • Audit log retention: 365 days
  • Shared dashboard + team admin
  • Unlimited read actions

Enterprise

Custom plans for large organisations

Contact Sales from $1,700 / month
  • SAML SSO + RBAC
  • Custom SLA + dedicated support
  • Unlimited projects & writes
  • On-prem / private cloud deployment
  • Custom audit retention
  • Unlimited read actions
Get Started — start with Free

Upgrade to Pro or Team anytime from your dashboard.

All plans: $0.07 per write action over your monthly limit

Cancel anytime.

Get Started Now

Setup completes in minutes. No credit card required.

Get early access