CalypsoAI
Secure, orchestrate, and monitor AI across industries with CalypsoAI.. [Contact for Pricing]
Last verified:
What is CalypsoAI?
CalypsoAI (now rebranded as F5 AI Guardrails) is the leading enterprise AI security platform that provides comprehensive runtime security for deployed AI models, applications, and agents. The platform addresses data security, adversarial threats, and regulatory compliance by protecting AI systems from prompt injection attacks, jailbreaks, data leakage, and harmful outputs. It operates as a proxy layer between users and AI models, intercepting prompts before they reach the model and analyzing outputs before they return to users.
Key features include AI risk assessment for both public foundational models and in-house models, distributed data protection with real-time DLP, simplified compliance with automated auditing templates for GDPR, HIPAA, and EU AI Act, low-latency runtime security with dynamic model routing, content moderation filters for toxic/biased/inaccurate content, and model-agnostic functionality supporting popular enterprise and open-source AI. The platform also includes AI Red Team for continuous adversarial testing, customizable security scanners (both GenAI scanners using AI and keyword/regex scanners), prompt history logging, group-based access controls, and integrations with Slack and Microsoft Teams.
CalypsoAI is designed for enterprises and governments deploying generative AI at scale, including security teams (SecOps), platform engineers, data teams, and compliance officers who need to secure AI systems from pilot to production. It serves organizations that need to protect proprietary and sensitive data flows, enforce organizational access policies, maintain audit-ready observability, and ensure regulatory compliance while leveraging LLMs and AI agents.
CalypsoAI pricing
Pricing model: Freemium
Pricing is available on request only - CalypsoAI/F5 AI Guardrails does not publish public pricing on their website. There is no free tier or freemium plan available. Enterprise pricing is customized based on organizational needs, likely involving contact with a sales representative for quotes. The platform is positioned as an enterprise-grade solution for organizations deploying AI at scale, suggesting premium pricing aligned with enterprise security tools.
CalypsoAI pros
- Comprehensive runtime security for deployed AI models and agents
- Protects against prompt injection and jailbreak attacks
- Real-time data loss prevention for sensitive information
- Supports both custom/internal models and pre-configured providers
- Pre-configured integrations with OpenAI, Azure, AI21, Cohere, Anthropic, Groq, watsonx
- Automated compliance templates for GDPR, HIPAA, EU AI Act
- Low-latency protection without compromising performance
- Model-agnostic functionality for frontier and open-source models
- AI Red Team for continuous adversarial testing at scale
- Customizable scanners including GenAI and keyword/regex types
- Full prompt history logging with admin visibility
- Group-based access controls and team management
- Audit-ready observability with scanning and logging tools
- Dynamic model routing to avoid failover states
- Content moderation filters for toxic, biased, inaccurate content
- Embeddable chatbots for Slack and Microsoft Teams
- Human-in-the-loop intervention logic without alert deluge
CalypsoAI cons
- Only English language specifically supported for scanners
- Personal Health Information (PHI) not covered by scanners
- Admin dashboard cannot be customized
- Microsoft Copilot application not integrable (no API access)
- SaaS environment requires specific encryption setup
- Pricing available only on request (no transparent pricing)
- No free tier or freemium plan available
- Training datasets written only in English limiting multilingual detection
Frequently asked questions about CalypsoAI
What languages does CalypsoAI support?
At this time, only English is specifically supported. Prompts can be sent in different languages, and if the LLM successfully responds to a prompt in another language, this is purely coincidental. All training datasets used for the scanners were written in English.
Can a single user be added to multiple teams?
Yes, individual users can be added to multiple teams, giving them access to a variety of models and scanner settings specific to their various roles within the organization.
Is Personal Health Information (PHI) covered by the scanners?
No, Personal Health Information is not covered by the scanners. Organizations handling PHI need to implement additional protections or custom scanner configurations.
Can scanners be configured to block specific terms or names?
Yes. In addition to scanning for offensive terms and toxic language, the Blocked Term policy allows specific words, terms, and names to be scanned for and blocked from entering the LLM.
Do admins have visibility into users' prompt history?
Yes, the Prompt History logs give full admin access into all user prompts, including scanner settings and outcome (blocked vs. passed), enabling complete monitoring of AI interactions.
Is integration with Microsoft Copilot supported?
CalypsoAI can integrate with any LLM accessible through an API, including OpenAI running on Azure Services. However, the Microsoft Copilot application doesn't provide API access, so it cannot be integrated with CalypsoAI at this time.
Can CalypsoAI be used with custom models?
Yes, CalypsoAI can support any model, including both custom and internal models, in addition to pre-configured providers for seamless integration.
Which pre-configured LLM providers are supported?
Pre-configured providers include OpenAI, Azure, AI21, Cohere, Anthropic, Groq, and watsonx. While CalypsoAI supports any model through API access, these providers allow for seamless model integration without custom configuration.
How is customer data handled in the SaaS environment?
Customer data is encrypted at rest using KMS keys for Aurora RDS Databases. Elasticache implements encryption in transit with TLS required. TimescaleDB data is encrypted both in transit and at rest. Both active databases and backups are encrypted.
Do admins have the ability to monitor chat prompt inputs without blocking?
Yes, the audit setting allows prompts to be input without being blocked so admins can monitor user behavior and inputs without interacting with the chat or affecting user experience.