Daxa, formerly known as Cloud Defense, provides cybersecurity services for General Artificial Intelligence (Gen AI) and enables enterprise Gen AI compliance and Data Access Control/Governance capabilities within AI Apps. Founded in 2019 and based in Cupertino, Ca., Daxa offers products including Pebblo and Proxima that can be used to protect enterprise data connections and workflow processes.
Daxa was founded in Cupertino, Ca. by Huseni Saboowala and Sridhar Ramaswamy. It was originally called Cloud Defense.
The Company received $3.13 million in funding in its Series A round, which was led by DreamIt Securetech.
In addition to raising capital, the Company rebranded itself from Cloud Defense to Daxa and focused on developing an AI Security Platform.
AWS-based with Kubernetes orchestration
Gen-AI application frameworks like Langchain for securing LLMs, agents, and datastores with policy guardrails
Inferred from company expertise statements on official site and product descriptions
| Service | Cost | Details | Source |
|---|---|---|---|
| Basic Tier | $500-2,000/month | Core monitoring metrics, limited retention | — |
| Business Tier | $2,000-8,000/month | Extended analytics, longer retention, custom alerts | — |
| Enterprise Tier | $8,000-25,000+/month | Full observability suite, extensive retention, advanced explainability tools | — |
| Per Agent Monitoring | $50-500/month per agent | Usage-based pricing depending on agent complexity | — |
| Feature | Daxa | Datadog AI | Arize AI | Scale AI |
|---|---|---|---|---|
| Core Monitoring | Yes | Yes | Yes | Yes |
| Behavioral Analysis | Yes | Partial | Yes | Partial |
| Explainability Tools | Yes | Yes | Yes | Yes |
| Runtime Security | Yes | No | Partial | No |
| EU AI Act Compliance | Yes | Partial | Partial | No |
| Starting Price | $500/mo | $500/mo | Custom | $2,000/mo |
| Free Tier | No | No | Limited | No |
| Enterprise SSO | Yes | Yes | Yes | Yes |
| API Access | Yes | Yes | Yes | Yes |
| SOC 2 Certified | Likely | Yes | Yes | Yes |
Daxa’s focus is on AI Governance and runtime security for agentic AI; whereas Datadog is focused on providing a broad range of infrastructure monitoring capabilities, using AI as part of its offerings. Therefore, Daxa will best meet your needs for AI compliance.
Use Daxa for AI agent security/governance; use Datadog for general observability.
Arize is focused on providing model observability for ML; whereas Daxa is focused on the runtime behavior of AI agents and their security. Also, Daxa provides enterprise-level access control capabilities that are stronger than those provided by Arize.
Use Arize for your ML models; and Daxa for your deployed AI agents.
Scale is focused on data labeling and evaluation; whereas Daxa is focused on providing ongoing monitoring of your deployed AI. This makes Scale the best choice for when your in the development phase; and Daxa the best choice when you move into production deployment.
Use Scale for AI training; and Daxa for your deployed AI agents.
Daxa provides the Pebblo MCP (Model Context Protocol) as an intelligent security layer between AI agents (such as OpenAI) and enterprise data. The Pebblo MCP provides real-time protection by enforcing security policies, redacting personally identifiable information (PII), blocking prompt injections and preventing other potentially dangerous operations. It maintains a record of all actions taken.
Pricing details are not publicly disclosed and are generally customized for each enterprise customer. Please contact Daxa Sales for a quote based on your number of agents, data sensitivity, and deployment size.
Daxa focuses on providing a data-aware MCP gateway with semantic intelligence for OpenAI agents. This approach emphasizes PII redaction and behavioral reasoning. Zenity and Obsidian provide broader discovery of agents across platforms, but they have less focus on real-time, protocol level security.
Yes, Daxa does provide field-level PII redaction, prompt injection defense, policy enforcement, and complete audit logs that block malicious input and allow legitimate workflow.
Yes, Daxa directly integrates with OpenAI Agent's SDK via Pebblo MCP. Agents can run the business logic and Daxa handles the security aspects, requiring little integration work.
Daxa provides product demo's and integration guides on their web site. Enterprise customers can also get assistance for managing policies and deploying via the Pebblo Dashboard.
There is no publically announced free trial available; however, there are video demos of sandbox environments that show how the real time security features work. Enterprise style trials will be provided by contacting the sales department.
Primarily focused on OpenAI Agents and MCP protocol; May require additional development for non-OpenAI frameworks. Best suited for enterprises with autonomous AI deployments which need to meet regulatory requirements.
Daxa uses Pebblo MCP to deliver its specialized Safe Agent platform that addresses key security vulnerabilities associated with the deployment of autonomous AI systems in real-time, providing both PII protection and threat blocking, as well as audit capabilities. As such, Daxa has positioned itself favorably for use by OpenAI agent users who are focused on compliance rather than on broad discovery.
Daxa is best suited for Enterprise customers who are securing their OpenAI autonomous agents with production-grade Data Protection & Governance.
Daxa specializes in providing securement of OpenAI agents utilizing Pebblo MCP with real-time PII Redaction, Prompt Injection Defense, Policy Enforcement, and Audit Trails — positioned as Agentic SASE with semantic intelligence; provides competitive advantage in AI Safety; strong focus on Enterprise Compliance for Cognitive Enterprises.
Fair - product details from official site, videos, and blogs; limited public info on pricing, full docs, or customer case studies as enterprise-focused private company.
Daxa collaborates with OpenAI on integrating Agents SDK into Pebblo MCP, demonstrated through jointly produced videos showcasing secure autonomous agent deployment.
The first release of the Pebblo MCP Gateway combines data-aware guardian agents that are aware of what they're protecting with protocol handlers to provide a way to protect agents through semantic security and behavioral assessments.
Showcased in an OpenAI x Daxa video about how to build trust into your autonomous agents, demonstrating real time threat blocking and enterprise-wide visibility.
Daxa pioneered agentic SASE for cognitive enterprises, focusing on dynamic policy orchestration and providing complete visibility into each and every activity performed by your agents.
Provides semantic understanding of agent communication, intent parsing and real-time behavioral assessment
Provides complete visibility into agent activities, data flows and system interactions for compliance and optimization purposes
Provides rich behavioral metadata including configuration, permissions, tool access, memory usage, and execution patterns
Analysis of Agent Behavior to Detect Malicious Activity in Real-Time and Continuously Evaluate Normal Baseline Operation
The Visualization of Graph Intelligence to Display Agent-to-Agent Connections, Shared Resources, and Blast Radius
Audit Logging and Compliance Tracking for All Communications Between MCP Agents
Controls at the Protocol Level Combined With Runtime Enforcement Mechanisms Across Connectivity and Agent Operations
Anomaly Detection Based on Machine Learning Identifying Deviations in Expected Agent Behavior Patterns
Detection Engines Powered by Artificial Intelligence That Fingerprint Agents and Monitor Intent Patterns for Malicious Activity
Establish Normal Operation Patterns for Each Operational Context to Detect Deviation in Real-Time
Real-Time Validation of Context Ensuring Agents Maintain Appropriate State and Permissions During Execution
Dynamic Security Orchestration Adapts to Agent Behavior, Data Sensitivity, and Enterprise Risk Posture
Indicators of Compromise are Identified Through Behavioral Reasoning Beyond Simple Pattern Matching
Guardian Agents Are Continually Improving Their Understanding of Enterprise Patterns and Evolving Threats
Anomalies and Deviations in Agent Behavior are Immediately Notified
Alerting is Context Aware and Adapts to Agent Behavior and Risk Assessment
Visibility is Provided for All Agent Activities, Data Flows, and Compliance Status
AI-Powered Detection of Malicious Activity Through Analysis of Intent and Behavior
In Real-Time, All Audit Requirements and Policy Adherence for Agents are Tracked
Executive Views of Agent Security Status and Potential Blast Radius
Seamless Integration with Existing Security Infrastructure Including SIEM and SOAR
