octovision: AMDT's OT Asset Intelligence Layer, Explained

octovision is AMDT’s enterprise visibility and vulnerability intelligence layer for operational technology. It takes the configuration data that octoplant already collects from PLCs, HMIs, robots, drives and network devices, and turns it into a single view across plants: what assets exist, which firmware and software they run, which known vulnerabilities apply, and which are ageing out of support. AMDT announced octovision on 6 October 2025 and frames the pair as “two layers, one source of truth”: octoplant is the operational control layer, octovision the intelligence layer above it.

This page explains how octovision works, what it depends on, what it does well and what is still roadmap as of October 2026. For the company and the rest of the portfolio, start at the AMDT hub.

What octovision is, and what it is not

It is a reporting and analysis layer, not a data collector. octovision does not talk to devices. AMDT describes it as working “without additional traffic to the shop floor or disruption to production”, using existing project data rather than network scanning or agents. Everything it knows about a managed asset comes from what octoplant uploaded during a backup job.

It is not an OT security monitoring platform. It does not watch network traffic, detect intrusions or raise alerts on anomalous behaviour. Those are the jobs of Claroty, Nozomi, Dragos and their peers, and octovision positions itself as a complement to them, not a substitute. The comparison is covered on octoplant vs network OT security.

It is AMDT’s answer to a question octoplant could not answer well on its own: across twelve plants and thousands of devices, where are we exposed, and what should we fix first? octoplant holds the facts per device. octovision aggregates them for the people who manage risk and budgets across sites.

How octovision depends on octoplant data

Coverage in octovision equals coverage in octoplant. If a plant has 400 controllers and octoplant backs up 250 of them, octovision knows the firmware of 250. The other 150 are invisible unless someone imports them. That makes octovision an honest mirror of backup discipline: a gap in the dashboard is usually a gap in the backup jobs.

The data is deep but only as fresh as the last job. Because the inventory is read from project and configuration data, it carries detail that a passive network sensor often has to infer: vendor, device type, firmware version, engineering software version and job status. AMDT calls this “bottom-up asset intelligence”. The flip side is timing. A module swapped on Tuesday shows up after the next successful upload, not the moment it is plugged in.

Unmanaged assets can be imported, and are labelled as such. The Summer Release 2026 added CSV import of external assets from discovery tools, maintenance systems or spreadsheets. AMDT states that octoplant-managed assets “have their configurations backed up, versioned and recoverable” and everything imported shows as unmanaged. The managed versus unmanaged ratio becomes a coverage metric in its own right. See octovision asset inventory for how the two inventory methods compare.

Measure it against a public yardstick. CISA and its partner agencies published Foundations for OT Cybersecurity: Asset Inventory Guidance in August 2025, listing 14 high-priority attributes for every OT asset. Manufacturer, model, role and criticality sit naturally in a configuration-based inventory; IP and MAC addresses, ports and services, protocols and user accounts are network and host facts that a discovery tool or a site survey usually supplies. Checking octovision’s fields against that list shows which columns you will have to import rather than read.

octovision features at a glance

The feature names below follow AMDT’s product page and the octoplant.com octovision page, as of October 2026.

Feature	What it does	What it changes for a plant
Multi-site visibility	One view of assets, components, jobs and CVEs across plants and regions	Answers “where do we run this firmware?” without a spreadsheet chase
Bottom-up asset intelligence	Inventory of vendor, type, firmware and software read from project data	Version-exact facts for patch and audit decisions
CVE matching and risk scoring	Matches known vulnerabilities to asset versions and prioritises them	A ranked backlog instead of a raw CVE list
Criticality classification	Tags assets by process criticality, need for protection and safety	Risk ranked by production impact, not CVSS alone
Lifecycle and obsolescence	Flags aging, outdated or unsupported assets	Input to a migration roadmap and capital plan
AI-supported guidance	Suggests mitigation steps for findings; can run on-prem	Faster first triage, with a human deciding
Backup and job visibility	Shows job status across sites	Exposes sites where backups quietly stopped
Dashboards and reports	Pre-configured and department-specific reporting	One version of the truth for engineering, security and management
Export and open API	CSV, JSON and API to SIEM, CMDB, ticketing and analytics	Feeds the systems IT and security already run

Backup and job visibility is the most underrated row. A central view of which sites have jobs failing is the cheapest way to find out that a plant has not had a good backup since its server was patched in the spring. That finding costs nothing to fix today and a great deal after a controller fails.

Multi-site visibility in practice

Consider an illustrative case. A manufacturer runs eight plants, each with its own controls team and its own conventions. A vendor advisory lands on a Friday for one controller family below a certain firmware revision. Without a central view, the corporate OT security lead emails eight plant engineers, gets five answers by Wednesday, and two of those are from memory.

With octovision, assuming octoplant covers those controllers, the same question is a filter: show every asset of this type below this firmware, by site, with its criticality class. The answer arrives in minutes and it is the version read from the device, not the one in a three-year-old drawing. The plant consequence is time: the window between disclosure and a decision shrinks from weeks to a day, and the decision is made on facts.

The vulnerability management page covers what happens next: prioritisation, false positives and the response options.

Deployment: on-prem now, cloud on the roadmap

This is the point where AMDT’s two websites disagree, and a buyer should resolve it before a purchase order, not after.

Source (checked October 2026)	What it says about deployment
amdt.com, About octovision	Currently deployed on-premises, with “cloud and SaaS deployment options planned for the near future”
octoplant.com, octovision	“On-prem, cloud, or SaaS”, and “supports flexible deployment models, including on-premise and cloud”

Our reading is that on-premises is the available model today and cloud is the direction, with the product site running ahead of the corporate one. For many plants on-prem is the preferred answer anyway, particularly in pharma, utilities and defence supply chains where OT data leaving the site triggers a security review. For a multi-site enterprise, on-prem raises a practical question: where does the central octovision server live, and how do eight plant networks reach it through the segmentation you already have? That is a network architecture decision as much as a software one, and it belongs in an OT network assessment if the zones and conduits are not already documented.

The Summer Release 2026

The octovision Summer Release (“Built to See”) is the largest update since launch. It added:

• External asset import by CSV from discovery tools, CMMS or spreadsheets
• A managed versus unmanaged coverage metric
• Criticality classification across three dimensions: process criticality, need for protection and safety
• False-positive management, so a CVE marked as not applicable to an asset stops resurfacing
• Automatic server-side CVE updates

Earlier, octoplant’s Spring Release 107 (7 April 2026) introduced the octovision Export API, Asset Lifecycle Management and the Companion AI chat. The AI Companion page covers the AI side.

Limits and open questions

• No octoplant, no octovision. It is not a standalone product. A site without octoplant gets imported, unmanaged rows and nothing behind them.
• Freshness depends on job schedules. Hourly, nightly or weekly jobs mean very different data ages. Ask how stale a row can be.
• Deployment model is unresolved in public. See the table above.
• No public pricing. AMDT publishes no prices for octoplant or octovision, and the licensing relationship between the two is not public.
• No public certified integrations with Claroty, Nozomi, Dragos or Armis, or a certified ServiceNow app, as of October 2026. Export by CSV, JSON and API is documented; the effort to map it into your CMDB is yours.
• CVE matching quality depends on product naming. Matching a firmware string to the right CPE entry is hard for every vendor in this space. How octovision handles ambiguous matches is not documented publicly.
• The product is young. Announced October 2025, with major features arriving in April and summer 2026. Expect change, and check the current release.

How to evaluate octovision

• What share of our devices does octoplant back up today, per site? That number is octovision’s ceiling.
• Which deployment models are generally available now, and which are roadmap with a date?
• How are CVEs matched to firmware strings, and how often does the server update its CVE data?
• Can we see the evidence behind a match before we act on it?
• How is criticality assigned, by whom, and is it versioned when it changes?
• What does the export schema look like, and has anyone mapped it to our CMDB?
• Can the AI features be disabled, and where does inference run?
• What does octovision cost on top of octoplant, and how is it licensed across sites?

If the honest answer to the first question is “we do not know”, the place to start is an inventory of what you have, which is the core of a plant systems assessment.

Deeper

• octovision Asset Inventory: configuration-based inventory versus network discovery, and why plants need both
• octovision Vulnerability Management: CVE matching, criticality and a prioritisation workflow
• Asset Lifecycle and Obsolescence: lifecycle data and how it feeds a migration roadmap
• AMDT AI Companion: the AI features, the “AI is optional” stance and the governance questions

Back to the AMDT hub, or across to octoplant, the layer octovision reads from, and OTSSM and integrations, where octovision’s export meets ServiceNow and the rest of the IT stack.