OT Remote Access Gets a Front Door, and Legacy Systems Stay in Service
Barry-Wehmiller Design Group and Kasm Technologies have announced a partnership that integrates Kasm for OT into DG360, BWDG’s customisable OT security managed services platform. The combined offering gives manufacturers control and auditability over who connects to production systems, which assets they can reach and how long that access lasts. Kasm contributes a browser-based platform that brokers access to vendor tools from a single front door, runs each session in a disposable containerised workspace, and applies data loss prevention, key logging, file transfer and screen capture controls, watermarking and kiosk mode inside the session.
It also provides what the release calls cyber-wrapping for legacy and end-of-life systems, keeping older operating systems and applications in service without exposing their vulnerabilities directly. DG360 rests on the Five ICS Cybersecurity Critical Controls and draws on more than 450 system integration professionals across 24/7 managed services, working alongside partners including Claroty, TxOne, Salvador Technologies, Elastic and Xiid. The release states that manufacturing is the most cyberattacked industry with an average cost of $4.88 million per incident, without naming the source. It gives no named customer, no deployment count, no pricing and no performance data.
Why remote access is the exposure
The release makes a case worth examining because it is narrower and better aimed than most OT security arguments, and it starts from the operational facts rather than from a threat model. It says remote access sits at the centre of the exposure, and gives three reasons that are all operational rather than hypothetical.
OEM contracts often require it, which means the machine builder needs a route into the plant to support the equipment it sold. Skilled labour is scarce, so the engineer who can diagnose a specific control system may be in another country. And sites are spread across regions, so the person with the answer is rarely in the building. Those three conditions are not going away, which means the question is not whether remote access exists but what shape it takes.
The critique of VPNs is the part that lands. A VPN grants network access, and once a session establishes, the user can reach anything the routing permits. An engineer called in to look at one human-machine interface does not need that; they need access to that interface, for that session, with a defined set of permitted actions. The gap between what a VPN grants and what the work requires is where the risk lives, and the release is right that granularity and audit trails are what these workflows actually demand.
The legacy problem cyber-wrapping answers
The most useful capability in this announcement is the one described in the fewest words. Cyber-wrapping keeps older operating systems and applications in service without exposing their vulnerabilities directly, and the release says it removes the pressure to rip and replace.
That addresses the central unsolved problem in operational technology, and it is worth stating plainly why. A plant floor runs machines nobody can patch, because the operating system is out of support and the application was certified against it years ago. Nobody replaces it either, because the production line cost millions and works. Every security framework written for information technology assumes a patch cycle, and a plant floor frequently has none. The result is that the most common advice, patch or replace, is advice a plant manager cannot follow.
Wrapping the system rather than fixing it is the pragmatic answer, and it is the position that makes the most sense of OT security as plants actually run it. If the machine stays vulnerable but the only way to reach it is through a controlled session with logging and content restrictions, then the vulnerability is no longer reachable from the internet and the risk is materially reduced without touching the machine. That is a different security model from hardening the endpoint, and for end-of-life industrial equipment it is often the only one available.
Disposable sessions invert the trust assumption
The second design choice is subtler and more consequential than it first appears. Each session runs in a containerised workspace, and the release states that a compromised credential or endpoint therefore has no direct path to the plant floor.
Traditional remote access assumes the endpoint is trustworthy and the network is not, which is why it spends its effort on encryption and authentication. That assumption is wrong in practice. An engineer’s laptop is the most likely compromised device in the chain, and once an attacker holds it, a VPN hands them the network. Containerising the session inverts the model: the design treats the endpoint as untrusted, the workspace is disposable, and what reaches the plant floor is a rendered stream rather than a network connection.
That inversion also changes what an incident costs. A stolen credential in a VPN model means an attacker inside the network until someone notices and revokes. A stolen credential in a containerised model means an attacker inside a session that ends, on infrastructure that is rebuilt, with the plant floor never having seen their machine.
Controls inside the session are the operational detail
Most remote access products stop at authentication. They establish who you are and then get out of the way. The list of controls here operates after that point, and each answers a specific worry a plant manager has about letting an outsider in.
Data loss prevention and file transfer controls address exfiltration, which matters because a process recipe or a proprietary formulation is often the most valuable thing on the machine. Screen capture controls and watermarking address the simplest leak of all, which is someone photographing a screen or taking a screenshot. Key logging builds the audit record of what commands went in. And kiosk mode restricts the session to the application in question, so the user cannot wander from the human-machine interface into the rest of the system.
Taken together those describe a session with limits on what it can see and what it can take, and a record of what it did. That is a different proposition from remote access as a connectivity product, and it is closer to what a regulated manufacturer actually needs to sign off. OT security has spent a decade telling plants to segment their networks; this tells them what to do about the machines they cannot segment away from.
The ownership position is commercial, not technical
One line in the release deserves separating because it is a business decision rather than a capability. It states that the customer owns the access layer, so underlying tools can change over time while the access model stays in place.
That is the opposite of the usual arrangement, in which the access product is tied to the tooling beneath it and replacing one forces replacing both. For a plant that expects to run for twenty years and will change its security vendors several times in that period, an access layer that survives a vendor change is worth more than any single feature. It is also a position a managed service provider can take because its revenue comes from the service rather than the licence, and it makes the integration partner a plausible long-term custodian of the front door.
Why the front door matters more than the tool
One architectural consequence deserves separating because the release states it in passing and it changes how OT security is bought.
A plant running three vendors runs three remote access products, three sets of credentials, three audit logs and three patching obligations. Each vendor arrives with its own way in, and the plant ends up with a perimeter made of whatever each supplier installed. The single front door replaces that with one place where identity, authorisation and the record live, which is the only arrangement in which an OT security posture is auditable at all.
That is also why the partner roster in this release matters more than it would in a general announcement. Claroty, TxOne, Salvador, Elastic and Xiid each occupy a different part of the OT security stack, and an access layer that sits above all of them rather than beside one of them is the thing a plant with an existing estate actually needs.
What would complete the picture
Four additions would let a plant engineer evaluate this rather than admire it.
A named customer is the first. A combined offering from two vendors with a launch at a trade fair is a proposition, and one manufacturer running it in production would establish whether it holds under real conditions.
Deployment mechanics are the second and the most practical. How does a containerised session reach a machine on an isolated plant network? That requires something on site, whether a gateway, a connector or an appliance, and the release describes a multi-tenant software-defined architecture without saying what sits at the edge.
Session performance is the third. Rendering a streamed workspace over a link to an engineer in another country is a different experience from a local session, and the latency a technician should expect determines whether the tool is usable for interactive diagnostics or only for observation.
And pricing is the fourth, including whether Kasm licenses the access layer per user, per session or per site, because that determines whether the model works for a manufacturer with forty small plants as well as one with four large ones.

What to watch
Whether the companies name a manufacturer running the combined offering in production, because that is what turns a partnership into evidence. Whether the release describes the edge component, since the architecture at the plant boundary decides how deployable this is. And, whether the cyber-wrapping claim survives scrutiny on genuinely end-of-life systems, because wrapping a machine that speaks an unauthenticated industrial protocol is harder than wrapping a modern application. And whether the access layer outlives its first underlying tool, because the ownership promise is the most distinctive commercial claim here and the easiest to test over time.

