# CISO or buyer

> Ten minutes, no install: what Keryn tells you about an AI product, how it maps to the frameworks you report on, what is measured and what is not, and what it costs.

Source: https://keryn.dev/docs/start/security-leader/

**You get:** a release decision you can explain, a mapping of the findings to OWASP LLM, NIST AI RMF, ISO/IEC 42001, the EU AI Act and five more, the start of a posture assessment, and the facts behind a buying decision. **You need:** a current browser. Nothing is installed and no account is needed.

## 1. A result on real code (2 minutes)

Download [helpdesk.zip](https://keryn.dev/samples/helpdesk.zip), a small AI help-desk agent. Open [keryn.app](https://keryn.app), choose **File** under **Scan** and drop the zip. The scan runs in your browser tab; the code is not uploaded.

Keryn Web opens the release decision: **Policy blocked (Hold)**. Two findings hold it, each with a traced flow from the user's prompt to a shell. That sentence is what an engineer and an auditor can both check: the decision, the rule, and the evidence behind it. The [AppSec engineer path](https://keryn.dev/docs/start/appsec-engineer/) walks through the evidence.

## 2. Map it to your frameworks (3 minutes)

Open **Controls**:

![The Controls view of Keryn Web for helpdesk.zip: 5 of 10 OWASP LLM categories have findings, with tabs for OWASP Agentic, ATLAS, NIST AI RMF, ISO 42001, the EU AI Act, NIST 800-53, ISO 27001 and CWE.](https://keryn.dev/screens/role-leader-controls-1280-light.webp)

*Controls for helpdesk.zip: 5 of 10 OWASP LLM categories have findings, with a tab for each framework. Taken from keryn.app main-5g3Rr4q1.js.*

Five of the ten OWASP LLM categories have findings (LLM05 Improper Output Handling with a cross-file flow, LLM09, LLM07, LLM10 and LLM06). The tabs give the same findings for OWASP Agentic, MITRE ATLAS, NIST AI RMF, ISO/IEC 42001, the EU AI Act, NIST SP 800-53, ISO/IEC 27001 and CWE. These are references for a reader, not a conformance or certification claim, and the page says so.

## 3. Start a posture assessment (3 minutes)

Choose **Posture** under **Settings**, or go to `keryn.app/#/posture`:

![The posture assessment of Keryn Web after the helpdesk.zip scan: the Quick, Standard and Full depths, the overall score, maturity by domain, and answers suggested by the scan.](https://keryn.dev/screens/role-leader-posture-1280-light.webp)

*The posture assessment after the scan: depths, the overall score, maturity by domain, and answers suggested by the scan. Taken from keryn.app main-5g3Rr4q1.js.*

The assessment asks about practices a scan cannot see (ownership, review, incident response) at three depths: Quick, Standard and Full. The scan suggests answers where it has evidence; confirm only what you know is true. The scores stay at zero until questions are answered, because nothing is assumed.

## 4. What is measured, and what it costs (2 minutes)

- **[Evaluations](https://keryn.dev/evals/)**: Keryn against reference tools and labelled corpora, with the misses and the cases it does not handle kept in the results.
- **[What Keryn does not do](https://keryn.dev/docs/concepts/limits/)**: the limits, stated up front.
- **[Privacy](https://keryn.dev/privacy/)**: every network call the software can make. Local scans send nothing.
- **[Plans and prices](https://keryn.dev/pricing/)**: local scanning is free; paid plans for Keryn Cloud are not open yet.
- **[Single sign-on and directory sync](https://keryn.dev/docs/cloud/enterprise-sso/)**: how Keryn Cloud connects to your identity provider, removes leavers through SCIM, and what each role may do.
- **[Download](https://keryn.dev/download/)**: what is available today (Keryn Web) and what is in pre-release.

## Next

- [Controls, posture and hand-off reports](https://keryn.dev/docs/guides/controls-and-posture/).
- [Design partners and pilots](https://keryn.dev/partners/design-partner/), if you want to run Keryn on your own systems with us.
