← Learn
Definition8 min read

On-Premise Vulnerability Assessment: Scanners, Air-Gapped Options and Exploit Validation (2026)

Published by Zero Hunt, an autonomous AI red team on an on-premise appliance running private AI: automated penetration testing for networks and infrastructure, black-box or gray-box, with a human approving every step that matters.

Short definition

An on-premise vulnerability assessment runs the scanner, the console and the findings database on hardware you control, so the list of your weaknesses never sits in a vendor's cloud. Most vulnerability management products are now cloud platforms with local scanners; only some keep everything inside, and fewer still prove which findings an attacker can actually exploit.

Why this matters now

A vulnerability database is a ready-made attack plan: every unpatched host, its version and its severity in one place. Defense, government, banks, hospitals, utilities and OT operators often cannot let that data leave their network, and isolated networks cannot reach a cloud console at all. A scan also stops at detection: NIST SP 800-115 warns that scanners can have a high false positive rate, so knowing which findings are really exploitable is the second half of the job.

Key points

  • ▸Ask two questions first: where do the console and the scan results live, and does the product keep working with no internet connection?
  • ▸Tenable Security Center, the Rapid7 InsightVM and Nexpose Security Console, Greenbone OPENVAS appliances and Fortra VM On-Premises run on your hardware. Tenable, Rapid7 and Greenbone document offline or air-gapped updates.
  • ▸Qualys VMDR is a cloud platform with local scanner appliances; Qualys offers an offline scanner for air-gapped networks and a separate on-premises Private Cloud Platform.
  • ▸Scanners detect and prioritize. They match versions and checks against known vulnerabilities and do not prove that an attacker can use a finding in your network.
  • ▸Exploit validation closes that gap. Run it on premise too, or the proof of what is exploitable leaves your network after the inventory did not.
  • ▸Vendor documentation changes. Treat this page as a shortlist and confirm every point in writing.

What on-premise means for a vulnerability assessment

Vendors use the word for three different architectures. Ask which one you are buying:

  • Local scanner, cloud console

    Scanner appliances or agents run in your network, but results are uploaded to the vendor's platform, where you view, prioritize and report. This is how most current vulnerability management platforms work.

  • Self-hosted console with cloud features

    The console and the database run on your server, but some features (dashboards, analytics, threat feeds) still need the vendor's cloud.

  • Self-contained

    Scanners, console, database and updates all work on hardware you control, including with no internet connection. Updates arrive as files you move across the air gap.

Only the third keeps your full inventory of weaknesses inside your perimeter at all times. For a vulnerability assessment the question is not only where the scanner runs, but where the results are stored and who can read them.

On-premise vulnerability assessment options at a glance (November 2026)

Console and results on your hardware

Tenable Security Center
On-premises software or Tenable Core image
Rapid7 InsightVM / Nexpose
Security Console on premise; dashboards and remediation projects use the cloud Insight Platform
Qualys VMDR
Cloud platform with local scanners; on-prem Private Cloud Platform sold separately
Greenbone OPENVAS SCAN
Physical and virtual appliances
Fortra VM On-Premises
On-premises edition
Zero Hunt
Appliance with its own AI models

Works air-gapped

Tenable Security Center
Documented offline plugin and feed updates
Rapid7 InsightVM / Nexpose
Documented offline update bundles
Qualys VMDR
Offline Scanner Appliance documented; not documented for the Private Cloud Platform
Greenbone OPENVAS SCAN
Airgap mode, updates by USB or data diode
Fortra VM On-Premises
Not documented
Zero Hunt
Documented, black-box only in that mode

Proves exploitability

Tenable Security Center
Detects and prioritizes (exploit maturity as a risk input)
Rapid7 InsightVM / Nexpose
Exploit Exposure links findings to Metasploit modules for verification
Qualys VMDR
Not documented
Greenbone OPENVAS SCAN
Vulnerability tests only
Fortra VM On-Premises
Identifies, assesses and prioritizes
Zero Hunt
Attempts exploitation and records signed proof

Sources: vendor product pages and documentation, checked 17 November 2026. "Not documented" means we found no statement either way. Zero Hunt is an autonomous pentesting product, not a vulnerability scanner replacement: see the last sections.

Scanner products you can run on premise

  • Tenable Security Center

    (named Tenable.sc until 2023). Tenable's on-premises vulnerability management; Tenable states that your data remains on-premises. It installs as software or as a Tenable Core image (VMware, Hyper-V, KVM or bare metal) with Nessus scanners, and its documentation describes offline plugin and feed updates for air-gapped networks, with one Security Center and its scanners inside each isolated network. Tenable Vulnerability Management is the cloud product.

  • Rapid7 InsightVM and Nexpose

    The Security Console is an on-premises scanner and management system on Linux or Windows. Offline update bundles exist for consoles with no internet access. Dashboards and remediation projects need the cloud Insight Platform, and InsightVM is now sold as part of Rapid7 Exposure Command.

  • Qualys VMDR

    A cloud platform with scanner appliances in your network. The Offline Scanner Appliance scans air-gapped networks without contacting the Qualys platform. Qualys also sells the Private Cloud Platform, a 2U on-premises hardware platform with the features of its cloud, updated and maintained by Qualys.

  • Greenbone OPENVAS SCAN

    Physical and virtual appliances built on OpenVAS. An Airgap mode separates a connected master from an isolated sensor, with updates moved by USB drive or through an (S)FTP server acting as a data diode. The open-source OpenVAS scanner can also be self-hosted, with the community feed.

  • Fortra VM On-Premises

    The on-premises edition of Fortra Vulnerability Management (formerly Frontline VM), to identify, assess, prioritize and track vulnerabilities. We found no documented air-gapped mode.

What a vulnerability scan cannot tell you

A scanner identifies hosts, services and versions and matches them against a database of known vulnerabilities. NIST SP 800-115 is direct about the limits: vulnerability scanners can have a high false positive rate, and while scanners check only for the possible existence of a vulnerability, the attack phase of a penetration test exploits the vulnerability to confirm its existence. Scanners also miss combinations: several low findings can add up to a path to domain admin that no single severity score shows.

The result is a long list ranked by CVSS or by the vendor's risk score, where most items are not exploitable in your environment and a few chained ones are. That gap is what exploit validation closes, and it is the difference between a vulnerability assessment and a penetration test, explained in VA vs PT.

Automated vulnerability assessment with exploit validation

Automated and autonomous penetration testing adds the missing step at the frequency of a scan. The tool discovers assets, matches versions to known vulnerabilities like a scanner does, then tries to exploit them, chain them and move further, and records proof for each finding it confirms. AI-driven tools choose the next step from what they find instead of following a fixed playbook.

This is not a replacement for your scanner. A scanner still gives the widest coverage of known weaknesses and patch levels; an automated pentest tells you which of them matter first. The two work best together: scan everything, validate what is exposed, retest after the fix.

On premise, the same rule applies to both layers. An exploit-validation tool learns even more than a scanner: credentials, reachable paths and working proofs of exploit. If its AI model or console runs in a vendor cloud, that data leaves your network. The list of on-premise AI pentesting tools shows where each product runs its model and whether it works offline.

How to choose

  1. Decide whether scan results may leave your network. If not, rule out cloud consoles, even with local scanners.
  2. Decide whether the tool must work with no internet connection, and ask how plugin and feed updates cross the air gap and how they are verified.
  3. Ask which features stop working offline: dashboards, threat intelligence, exploit and risk feeds, reporting.
  4. Check the scope: network and hosts, web applications, cloud, OT. On-premise editions sometimes lack modules the cloud edition has.
  5. Ask whether findings are only detected or also validated, and how a validated finding is proven.
  6. If you add exploit validation, apply the same questions to it: where the AI model runs, who approves risky actions, where evidence is stored.
  7. Check what your rules ask for: PCI DSS needs both scans and penetration tests, and Italy's ACN measures accept vulnerability assessment and/or penetration testing. See requirements by regulation.
  8. Put every answer in the contract; the buyer's guide to on-premise pentest platforms has the questions to send.

Zero Hunt's approach (vendor section)

The sections above apply whatever product you choose. This one describes how Zero Hunt, the product behind this site, fits.

  • Assessment and proof in one campaign. Zero Hunt is an autonomous AI red team for networks and infrastructure. Its vulnerability analyst agent matches service banners and versions against CPE and NVD and cross-references CISA KEV and EPSS; exploit and post-exploitation agents then try to prove each finding, so every finding records whether it was exploitable in your environment.
  • On-premise, private AI. It runs on an appliance on your site with its own ZeroHunt Apex models and no external AI API; no customer data leaves it. Air-gapped mode is documented: updates arrive as signed offline bundles, and public OSINT and source downloads are switched off, so tests run black-box.
  • Black box and gray box. Campaigns run black box by default, and gray box with credentials you provide or through analysis of the source of the exact software version in use.
  • Human in the loop. Five autonomy levels decide which actions wait for an operator; exploit verification is enabled only at the two highest levels.
  • Evidence. Every attack attempt is an Ed25519-signed record in a SHA-256 hash chain, and findings map to 34 compliance frameworks.
  • Next to your scanner. Findings are exported through the API with CVE IDs and asset addresses, so you can correlate them with Tenable, Qualys or InsightVM results. There is no automatic import of scanner results.

It does not replace a full vulnerability scanner or patch-level inventory, a PCI ASV scan, or the accredited testers some rules require. Read about the on-premise AI red team.

Questions

Can vulnerability assessment run fully on premise? Yes. Tenable Security Center, the Rapid7 Security Console, Greenbone OPENVAS appliances and Fortra VM On-Premises run on your own hardware. Check which features still need the vendor's cloud: Rapid7 dashboards and remediation projects do.

Which vulnerability scanners work in an air-gapped network? From vendor documentation checked in November 2026: Tenable Security Center (offline plugin and feed updates), Rapid7 Security Console (offline update bundles), Greenbone (Airgap mode with USB or data-diode updates) and the Qualys Offline Scanner Appliance.

Is Qualys on-premise? Qualys VMDR is a cloud platform; its scanner appliances run in your network and send results to the platform. Qualys documents an offline scanner for air-gapped networks and sells a separate on-premises Private Cloud Platform.

Does a vulnerability scanner prove that a vulnerability is exploitable? No. Scanners detect and prioritize. NIST SP 800-115 says scanners check only for the possible existence of a vulnerability, while a penetration test exploits it to confirm it. Rapid7 links findings to Metasploit modules so you can verify them; automated pentesting tools attempt the exploit themselves.

What is an automated or AI vulnerability assessment tool? Usually one of two things: a scanner with automated scheduling and AI-assisted prioritization, or an autonomous pentesting tool that also attempts exploitation. Ask which one it is, and where its AI model runs.

Sources

Goes deeper

Want this against your environment?

Book a 30-minute scoping call — we will map this directly to your current compliance scope and threat profile.