Kastra Labs Inc.
Kastra provides a runtime authorization layer for AI systems, ensuring deterministic policy enforcement before AI actions are executed.
At a Glance
- Engineering organizations
- Fortune 100 banks
- Healthcare systems
- Federal agencies
AI Tools by Kastra Labs Inc.
(1)
Kastra
AI Agent Runtime Authorization Layer
Discussions
No discussions yet
Be the first to start a discussion about Kastra Labs Inc.
Latest News
Kastra Launches on Product Hunt: Runtime authorization for AI Agents
Show HN: Policy enforcement for Claude Code, Cursor, and Codex
Kastra Releases Comprehensive Guide to AI Agent Governance
Technical Deep Dive: Achieving Sub-Millisecond Policy Evaluation
Products & Services
Every model action (tool call, prompt, API request) checked against policy before execution.
Local governance daemon for developer machines, supporting Claude Code, Cursor, and Codex CLI.
Scanning tool for AI agent history to identify risky actions and auto-draft safety policies.
Hash-chained audit trail providing cryptographic evidence for every AI decision.
Market Position
Positions as the 'AWS IAM for AI', focusing on deterministic runtime authorization rather than post-hoc monitoring or prompt filtering.
Leadership
Founders
Carlos Jimenez
Founder of Kastra Labs. Previously active as a product-focused 'maker' and developer. Co-built Kastra after an AI agent nearly deleted a production database.
Fernando Vintacourt
Co-founder and Software Engineer at Kastra Labs. Previously a Software Engineer at ZSix, Nexos Creative, and sarabitu. Alumnus of Universidade Federal do Rio Grande do Sul (UFRGS).
Executive Team
Carlos Jimenez
Co-Founder
Product and growth lead at Kastra Labs.
Fernando Vintacourt
Co-Founder
Technical lead and software engineer at Kastra Labs.
Founding Story
The founders built Kastra after a coding agent (Cursor) almost executed a destructive 'DELETE' command on a production database. They realized AI needs a deterministic safety layer that operates outside the probabilistic model.
Business Model
Revenue Model
Subscription-based model with tiered plans based on seats and authorization volume.
Pricing Tiers
1 developer, 50K decisions/mo, 7-day audit retention.
1 developer, 1M decisions/mo, 90-day retention, custom policies.
Pooled infrastructure, 10M decisions/mo, 1-year retention, shared audits.
Dedicated VPC, single-tenant, air-gap support, 15-min SLA.
Target Markets
- Engineering organizations
- Fortune 100 banks
- Healthcare systems
- Federal agencies
- Coding agents security
- Autonomous enterprise workflows
- HIPAA-compliant AI diagnostics
- Regulated industry AI adoption