Lyzr Launches Opencontroller in AWS Marketplace, So Enterprises Can Own What Acts on Their Behalf
News originally published by
EIN PresswireOne control plane for every AI agent, on any cloud, running inside the customer's own AWS account.
NEW JERSEY, NJ, UNITED STATES, November 19, 2026 /EINPresswire.com/ -- Lyzr, an enterprise AI agent platform provider, today announced that Opencontroller, the control plane it launched in September, is available in AWS Marketplace. Opencontroller finds every AI agent an enterprise runs, gives each one an owner, and controls what it can call, spend and touch. It runs on Amazon EKS inside the customer's own AWS account, works with the agents already running on Amazon Bedrock AgentCore and other platforms, and can be bought through an AWS Marketplace private offer that can count toward the customer's AWS commitment.Enterprises now build agents in many places at once: Amazon Bedrock AgentCore, Microsoft Copilot Studio and Foundry, Google Cloud, Databricks, Snowflake and developers' laptops. Each platform has its own console, its own keys and its own bill. When the board asks what agents the company has, what they can do, what they cost and who can stop them, most CIOs and CISOs answer from a spreadsheet, or not at all. IBM's 2025 Cost of a Data Breach Report found that one in five organizations reported a breach caused by shadow AI, and that 97% of organizations whose AI models or apps were breached had no AI access controls.
Opencontroller answers those four questions from one console, without moving a single agent.
1. Find: A read-only scan of the customer's clouds, clusters and developer laptops builds one live inventory of every agent and everything it uses: models, tools, MCP servers, identities and keys. It never invokes an agent, reads a conversation or fetches a secret.
2. Ship: Every agent gets a named owner, an approval and its own workload identity before it reaches production, so teams can retire the shared API key.
3. Run: An agent gateway decides who can call each agent. A model gateway puts every model call on a budget and runs guardrails that block personal data, secrets and prompt injection before the call lands. When an agent goes wrong, its owner can block it on the platform it runs on, with a name and a reason recorded for every change.
4. Improve: Production failures are grouped and turned into test cases, so each release is better than the last.
Opencontroller Runs Inside the Customer's AWS Account
Opencontroller runs where the customer's data already lives. Lyzr ships it as Helm charts that deploy on Amazon EKS in private subnets. It uses Amazon Bedrock and Bedrock Guardrails, keeps secrets in AWS Secrets Manager, and encrypts every store with the customer's AWS KMS keys. Prompts, keys and telemetry never need to reach Lyzr, and changes to AWS agents also land in AWS CloudTrail. For Amazon Bedrock AgentCore, Opencontroller reads runtimes, gateways, memories and policy engines with a read-only IAM role and checks them against the AWS Security Hub controls for AgentCore. It extends AgentCore across every other platform the enterprise uses.
“Boards had a mandate for Enterprises to adopt AI and deliver ROI. Today, CxOs are discovering that AI and agents are everywhere and no one can report where and how they are being used” said Akshay Patel, VP of AI Strategy at Lyzr AI. “Old governance and security tools are failing because AI Agents need governance rebuilt from agentic first mindset. It has to be sovereign and in full control of the enterprise. This is Opencontroller. It gives back control to CISOs, CIOs, and CAIOs while enabling the company to speed up responsible, trusted AI use that is auditable, measurable, and actionable with a click of a button. The ability to stop a rouge agent or capture the enterprise AI posture and report it back to the board at anytime is powerful.”
Lyzr Is Trusted by Enterprises Across Industries
Lyzr built Opencontroller with a Fortune 50 company as a design partner and customer one and Lyzr itself is customer two for our own engineering and AI operations. Today, Lyzr's customers run it fully air gapped and sovereign in their infrastructure choice to one control all clouds, enterprise SaaS application, and developer laptops without moving teams off their tools. Adoption and governance are immediate. Lyzr's has a history of serving large customers such as Prophet, Verifone, Nordstrom, Accenture, WTW, KPMG, and many others.
Opencontroller is available today in AWS Marketplace for CIOs, CISOs, CTOs and AI leaders who want one control plane for every agent they run. Getting started takes less than a week, and discovery is read-only, so it starts without changing anything the customer runs. Customers can ask their AWS account team, or email ashish.mali@lyzr.com for a private offer. Accepted offers appear on the customer's AWS invoice. To see a live discovery scan, book 30 minutes at lyzr.ai/opencontroller and bring the agent you worry about most. Lyzr is at AWS re:Invent, booth #134, November 30 to December 4.
About Lyzr
Lyzr AI is a full-stack enterprise AI platform that enables organizations to build, deploy, and govern AI agents for mission-critical operations. Designed for secure, production-grade deployments, the platform combines modular AI infrastructure with enterprise governance, observability, and deployment capabilities, allowing organizations to run AI entirely within customer-controlled environments. Headquartered in New York with an engineering hub in Bengaluru, India, Lyzr AI now has a presence across the United States, India, the United Kingdom, and Dubai, serving enterprises across banking, financial services, insurance, healthcare, and the public sector. Lyzr is a registered AWS Marketplace seller, is in AWS ISV Accelerate, and was a global finalist for GenAI Innovator Partner of the Year in the 2025 AWS Partner Awards. Lyzr holds SOC 2 Type II, ISO 27001, ISO 42001, HIPAA and GDPR.
Anirudh Narayan
Lyzr AI
Visit us on social media:
LinkedIn
Legal Disclaimer:
EIN Presswire provides this news content "as is" without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the author above.
