"As Ellavox grew to have nearly 1,000 AI agents in production, we realized we needed a powerful control plane to govern exactly how AI actually behaves - just like we needed Terraform's control plane to manage cloud deployments," said Sean Alsup, Ellavox's CEO. "We initially built ECP for ourselves, but after the recent McKinsey and Alibaba events where AI went very wrong, we decided to offer ECP to the entire industry now," he added.
Ungoverned AI Reaches A Breaking Point
An autonomous AI agent by security firm, CodeWall AI, recently breached an internal McKinsey AI platform called Lilli in under two hours, exposing 46+ million AI interactions, 728,000 files, 57,000 user accounts, and most critically, got write access to system prompts controlling AI behavior.
At an affiliate of Alibaba, an AI agent called ROME engaged in unauthorized cryptocurrency mining and covert network tunneling without human instruction, diverting resources, inflating costs and posing potentially serious legal and reputational risks.
Both issues exposed significant security, governance, and control lapses.
How ECP Would Have Prevented Both
ECP operates as a layer between an organization's AI applications and the AI models they access. It gives organizations control across the full lifecycle of an AI agent's deployment using:
- Immutable, Versioned Prompt Artifacts & Registries: Prompts become versioned, locked, and cryptographically governed artifacts stored in centralized registries, with full version history, change management workflows, and the ability to promote or roll back prompt versions across environments without redeployment.
- Policy Enforcement: Allows users to define and enforce rules governing how agents behave at runtime: content policies, output constraints, model routing rules, and access controls which get applied consistently across every AI agent call.
- Tool Access Control: Enables granular, role-based controls over which agents can invoke which tools, APIs, and external services with the ability to approve, restrict, or audit tool usage in real time.
- Runtime Governance: Delivers span-level observability into every agent interaction, including LLM-as-judge evaluation, statistical drift detection, and distributional analysis so teams know when behavior changes before users do.
- Auditability and Compliance: Provides complete, immutable audit trails of every prompt, policy decision, and tool call giving compliance, legal, and security teams the visibility they need to operate AI systems responsibly.
Availability
Elacity Control Plane (ECP) is available now for developers, service providers and enterprises at www.elacity.ai
About Ellavox AI
Ellavox AI (www.ellavox.ai) is an innovative provider of AI tools, and AI Voice Workers for Logistics, Multi-family Apartments, and Customer Service.
Media Contact
Rich Waidmann
rwaidmann@elacity.ai
Logo - https://mma.prnewswire.com/media/2949941/logo_merged_Dec_2025_Logo.jpg
View original content:https://www.prnewswire.co.uk/news-releases/ellavox-launches-groundbreaking-elacity-control-plane-for-ai-systems--first-of-its-kind-platform-that-secures-governs-and-audits-ai-systems-at-scale-302734126.html
