Ethyca Launches Astralis for Real-Time AI Governance
Ethyca launched Astralis on August 4, a platform for governing enterprise AI models' and agents' use of company data in real time. According to Ethyca and SiliconANGLE, the product combines automated privacy and AI assessments with purpose-based access controls that evaluate data requests as they arrive. Ethyca said Astralis runs within a customer's cloud environment.
Ethyca launched Astralis on August 4, introducing a platform designed to govern how enterprise AI models and agents use company data in real time. SiliconANGLE reports that the offering combines automated assessment workflows with controls that screen incoming data requests, while Ethyca describes the product as runtime governance for enterprise AI.
The launch addresses a practical governance problem created when agentic systems access data across multiple enterprise systems faster than privacy, security, and compliance teams can review individual actions. In its announcement, Ethyca argues that periodic reports, data inventories, and point-in-time assessments cannot keep pace with machine-to-machine interactions.
Two components for assessment and access control
According to SiliconANGLE, Astralis has two principal components:
- •A Large Language Regulatory Model for automating privacy and AI assessments. Ethyca told SiliconANGLE that work previously requiring 40 to 60 hours per assessment can take 20 to 40 minutes with the product.
- •A Purpose-Based Access Control engine that evaluates data requests at runtime and ties access grants to the approved purpose for using the data.
Ethyca's product announcement describes the central implementation premise as encoding organizational rules so systems can apply them when data is used. SiliconANGLE reports that the platform also tracks regulatory developments and requests additional information from internal experts when needed.
Deployment and scale claims
Astralis runs inside a customer's own cloud environment, according to SiliconANGLE, which reported that Ethyca said sensitive data does not leave the organization's control. The publication also reported Ethyca's claim that a large U.S. financial institution is using the platform to automate 6,000 requests per second, representing hundreds of millions of governed data decisions each month.
Ethyca founder and CEO Cillian Kieran told SiliconANGLE: "Organizations are not ready for the AI onslaught. Risk teams managing thousands of requests today will soon face millions. This is holding up AI adoption."
For ML and platform teams, runtime data governance is distinct from model-level evaluation or pre-deployment approval. Comparable enterprise approaches commonly require controls that can associate an agent's request with identity, authorization scope, data classification, and intended purpose at inference or tool-use time. The engineering question is whether those policy checks can be integrated into agent and data-access paths without creating unacceptable latency, availability, or observability gaps.
Ethyca counts The New York Times, Ramp, and Advance, Conde Nast's parent company, among its customers, according to SiliconANGLE. The available reporting does not provide independent performance benchmarks for Astralis, details of supported agent frameworks, or specifications for how its regulatory model reaches assessment conclusions.
Key Points
- 1Ethyca launched Astralis for real-time governance of enterprise AI data use, combining automated assessments with purpose-based access checks.
- 2Ethyca told SiliconANGLE its assessment automation reduces work from 40-60 hours to 20-40 minutes, a claim requiring deployment-specific validation.
- 3Comparable runtime-governance systems place policy evaluation on agent data-access paths, making latency, reliability, and auditability central engineering considerations.
Scoring Rationale
Astralis targets a material operational barrier to enterprise agent deployment: enforcing data-use policies at runtime rather than through periodic reviews. The product is relevant to teams building agents with broad data access, though the available evidence relies substantially on company claims and does not include independent technical benchmarks.
Sources
Primary source and supporting public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
