Culvii provides a governance-first TypeScript/Python SDK and managed service that helps enterprises build and operate AI agents with cryptographic identity, capability controls, and immutable audit trails. The platform enables security teams to approve and monitor agent deployments through features like human-in-the-loop checkpoints and multi-tenant key management. Culvii targets Fortune 500 companies, cybersecurity agencies, and manufacturers transitioning to agent-native operations.
Funding
Funding not disclosed
Founders
Product
Problem
Enterprises deploying AI agents face a fundamental tension: engineering teams want to ship quickly, while security teams need assurance that agents operate within defined boundaries. Traditional agent frameworks lack built-in identity, governance, and audit mechanisms, forcing organizations to build these controls from scratch or risk deploying agents with excessive permissions and no verifiable accountability.
Solution
Culvii provides a governance-first SDK and productized managed service that makes AI agents enterprise-ready. The TypeScript and Python SDK abstracts away cryptographic identity, capability governance, human-in-the-loop controls, and immutable audit trails, so engineering teams can ship agents that security teams will approve. Every agent receives a DID backed by Ed25519 keypairs, and every action is signed and recorded in a hash-chained Merkle Tree audit log. The managed service adds experienced operators—including Enterprise Operations consultants, AgentOps Engineers, and Runtime Experience Engineers—who handle deployment, incident response, and reliability with a 24-hour turnaround SLA. Together, these offerings give organizations a clear path from human-operated to agent-native operations.
Target Audience
Primary customers are Fortune 500 companies, cybersecurity agencies, and manufacturers that need to deploy AI agents in production with enterprise-grade security, governance, and auditability.
Features
- Cryptographic agent identity using DID and Ed25519 keypairs, with verifiable and revocable identities across tenants
- Capability registry with explicit grants, scoped permissions, expiration, instant revocation, and no ambient authority
- Immutable audit trails built on hash-chain buffers and Merkle Trees for tamper-evident, cryptographically verifiable logs
- Human-in-the-loop checkpoints that suspend execution and await approval via Slack, email, or webhook channels
- Multi-tenant key management with per-tenant scoping, rotation without downtime, and per-key rate limiting
- Provider-agnostic model configuration supporting OpenAI, Anthropic, Azure, or local LLMs
- Managed service tiers (Small Business and Enterprise) with 24-hour turnaround SLA and a graduation path to self-serve operation