EDAMAME on AWS Marketplace: AI Agent Security Through Your AWS Account

Thierry Messmer

AI agents now do real work inside the company. They read and write code, hold credentials and act on production systems, on employees' workstations, on CI/CD runners and on servers in the cloud. Many leadership teams have already made the first decision, which was to let them. The second decision is the one that now sits with the business as much as with security: who checks what those agents actually do with the access they were given?

EDAMAME answers that question, and large companies can now buy it the way they already buy their cloud. The EDAMAME Enterprise subscription is available on AWS Marketplace, launched with the support of Benilar.

What EDAMAME gives the business

EDAMAME is the trust layer for AI agents and the machines they run on. It watches every agent from outside the agent, with nothing installed inside it, and compares three things: what the human authorized, what the agent declared and what the machine actually did. That comparison turns into evidence the business can act on.

Know which agents you are running. EDAMAME finds every AI agent on workstations, CI/CD runners and servers, including the ones nobody declared, along with the MCP servers and tools each one reaches and the damage it could do if it were compromised.
Evidence instead of assumptions. When an agent drifts from what it was asked to do, or behaves like an attack (credential harvesting, token exfiltration, tool poisoning, supply-chain compromise), EDAMAME records it with the proof. Each AI check carries its OWASP GenAI, MITRE ATLAS, ISO/IEC 42001 and ISO 27001 references, so an auditor reads the result in the vocabulary of the standard.
Control through the tools you already own. When a machine stops deserving trust, EDAMAME tells the access controls you already run, including Netskope, Microsoft Entra, Google, GitHub, GitLab, FortiGate, Tailscale and NetBird, and they close access to code, secrets, production and other critical resources. Nothing is blocked on the workstation itself; on a CI/CD runner, the pipeline stops.

The business outcome is speed with control: autonomy is granted to agents step by step, against evidence, which is what lets a company say yes to more of them. How it works, agent by agent.

Why buying through AWS Marketplace matters

A new security category rarely stalls on technology. It stalls in procurement: a supplier to qualify, a contract to negotiate, an invoice for finance to route. In a Total Economic Impact study that AWS commissioned from Forrester Consulting in 2025, organizations buying through AWS Marketplace saw 60 percent faster procurement processes. For an AI agent security purchase, the difference shows up in two teams.

For finance

One bill. You pay AWS, and the charge lands in the same AWS account your company already uses to pay for AWS services, so the invoice sits in the AWS Billing and Cost Management console with the rest of your cloud spend (AWS Marketplace FAQ).
Spend you can allocate. Each AWS Marketplace purchase can carry its own purchase order number, which separates it from your infrastructure spend and assigns it to the right cost center.
A cost that does not grow with adoption. The Enterprise subscription is a flat price with unlimited users and unlimited AI agent deployments for the first year, so nobody has to count seats while agent use spreads across the company.

For procurement

A relationship you already have. The purchase runs through your existing AWS account and the AWS Marketplace buying process, instead of a new supplier relationship built from scratch.
Terms that fit your process. When your organization needs specific pricing or terms, EDAMAME prepares a private offer for the AWS account you designate, and you accept it in AWS Marketplace. A private-offer contract can also be billed as installments on a custom schedule instead of a single upfront payment.
Buying inside your own controls. Companies that curate approved software with Private Marketplace can add EDAMAME to their catalog, and teams that buy through Coupa or SAP Ariba can reach AWS Marketplace from those tools.

The Enterprise launch offer

The Enterprise subscription on AWS Marketplace covers:

12 months of AI agent security and AI governance;
EDAMAME Security for workstations, EDAMAME Posture for CI/CD runners and servers, including Amazon EC2, and EDAMAME Hub for fleet-wide governance and reporting;
unlimited users and unlimited AI agent deployments during the first year;
an onboarding session.

The price is USD 60,000 for the 12 months, paid upfront, excluding token costs.

Built on AWS, alongside AWS

EDAMAME runs as a multi-tenant SaaS platform on AWS, and its listing carries AWS's Deployed on AWS designation. It complements what AWS already secures. Under the AWS shared responsibility model, AWS secures the cloud itself, and what customers run on it, including how their applications and agents behave, remains the customer's responsibility. EDAMAME gives you the evidence for your side of that line: which agents exist across your estate, what each one can reach, whether it stayed inside what the human authorized, and whether the machine it runs on can still be trusted, on AWS and off it.

It also fits the stack you already run. EDAMAME complements your EDR, your security service edge and your identity provider rather than replacing them, and its findings reach the access controls you already use. That is how the Netskope integration works: EDAMAME applies a device tag while a device passes policy, and Netskope enforces your own access rules on it.

How to get started

Talk to us. A conversation with Thierry Messmer scopes the rollout with you and, if your procurement needs it, the terms of a private offer.
Subscribe with your AWS account from the EDAMAME listing on AWS Marketplace, or accept the private offer prepared for you. Your AWS administrator can confirm who in your organization is allowed to subscribe.
Onboard with our team. The onboarding session included in the subscription sets up EDAMAME Hub with you.

Book a conversation with Thierry, or see the 30-second overview: who checks what your AI agents do?