Kagliostro
Kagliostro.cloud provides an autonomous AI security agent that analyzes authorized code, applications, and APIs, then proves exploitability in a sandbox and prepares tested fixes for human approval. It targets SMEs without internal security teams and larger enterprises needing continuous remediation, with a 0% false positive rate and average fix delivery in 4.2 minutes.
- Cybersecurity
- AI Agents
- Artificial Intelligence
- Developer Tools
- Software Only
Funding
Founders
Founder details are not available yet.
Product
Problem
Security teams are overwhelmed by hundreds of untriaged alerts, often missing the one critical issue that matters. Traditional tools point out potential problems without verifying if they are actually reachable or dangerous, and they rarely prepare fixes, leaving vulnerabilities open for months. Annual security audits become obsolete quickly, creating noise that saturates developers while real risks remain exposed.
Solution
Kagliostro.cloud provides an autonomous security agent that takes a result-oriented mission rather than a predefined procedure. The agent analyzes authorized repositories and services, builds its own action plan, and determines what to examine and in what order. It investigates potential issues, proves exploitability in a sandbox environment, and prioritizes findings by explaining the urgency. Once a problem is confirmed, Kagliostro prepares the code fix, tests it to ensure no regressions, and presents the result for final approval. The entire loop—from objective to verified correction—runs autonomously, with the human retaining the last word before changes are applied.
Target Audience
Primary customers are small and medium-sized enterprises without an internal security team—including e-commerce, health, logistics, education, and NGO organizations—as well as technical teams and CISOs at larger enterprises needing continuous verified security remediation.
Features
- Fully autonomous security loop: finds, verifies, prioritizes, fixes, and validates without manual step-by-step guidance
- Sandbox-based proof of exploitability, eliminating false positives with a 0% false positive rate (PoC required)
- Automatic pull request generation with an average delivery time of 4.2 minutes
- Continuous 24/7 monitoring of 2,500+ repositories and APIs
- Objective-driven interface where users state goals like "secure our login service" rather than configuring tools
- Regression testing built into the correction workflow to ensure fixes do not break existing functionality
- Average detection time under 90 seconds from threat identification to alert
- Self-hosted deployment on the customer's own server with isolated per-client installations
- Ownership proof via DNS TXT record or well-known file, plus a digital mandate signed online
- Domain-verification process that checks every 10 seconds and refuses government and military sites
- French-language conversational interface explaining risks in plain language with urgency levels