Moonscape Business Solutions Tests OpenFrame for Co-Managed IT

Artem Rakhlev started Moonscape Business Solutions in May 2025, after 20 years in IT. Before that he did independent field engineering and ran a small MSP built mostly on hardware service calls. Moonscape runs co-managed IT: Artem's team of two works alongside his clients' own technicians, about 50 people in all.

He found OpenFrame while building his stack, through Flamingo's resource on open-source tools. At the interview in December 2025, he was piloting it on about five devices: his engineer's machine, a remotely managed door-access controller and a couple of his own devices. No end-user devices were on it yet.

His plan came with two conditions. Once end users can approve a remote session, or at least see one, and pricing is clear, he said: "I would try to push it across all my clients."

Artem started Moonscape as a step into the modern world, after years of hardware service calls. Building the stack came first, and the options split two ways. Big vendor suites had grown into monoliths that weren't feasible for a new company. Open-source tools were great when they worked, but keeping them running fell on him. In his words: "You want to service your clients, not your own infrastructure." His goal was to avoid overhead and not just pay for SaaS.

His model adds a twist. Moonscape does co-managed IT, which he calls "the best way to go." His team handles the menial tasks, and the client's own technician, who knows the business best, keeps the big picture. That means about 50 technicians need access: two of his own and the rest on the client side. Seat-based pricing meant paying for people who don't work for him. "Why should I be paying for them?"

Per-endpoint pricing leaves a different gap. He has seen security and monitoring tools bought for only part of a company, because covering everyone costs too much. "If it was deployed globally, you could find every single shadow IT practice or who's using unapproved AI and things of that nature. But it doesn't cover that simply because of costs."

Artem joined the OpenFrame beta and started small. The agent went on his engineer's machine, a couple of his own devices and a remotely managed door-access controller. That's about five devices, and none of them belong to end users yet.

Fae, the AI agent that sits on the client's machine, was the part he liked first: "I like the Fae concept." He tried it on his own laptop, asking whether updates were available. That run failed. Fae tried to do too much, and it never suggested opening Windows Update instead. What he wants there is a way for a technician to take over the chat, with a summary of what the AI already tried.

On the dashboard, the basics were in place: "You guys already have the active chats, resolve chats, how many devices online? All fine." Remote sessions worked on one device and showed a black screen on an Azure-registered one that had connected before. He hadn't spent much time with scripts, and he hadn't yet found a job for Mingo, OpenFrame's AI agent for technicians.

Asked what he'd hand Mingo, the list came quickly: clear cache and cookies, reset browser settings, check for Windows updates, kill a task, find devices over 90% RAM or under 10% free disk space, and install a script across his inventory.

At the interview in December 2025, Moonscape's results were about fit, not hours saved. OpenFrame ran on about five devices, including the door-access controller, and on no end-user machines.

Two things stood between the pilot and a rollout: "the security concern and the cost concern." The security one is specific. A technician can open a remote session without the end user seeing a prompt. For someone in accounting, or processing credit cards, Artem wants the user to approve the session, or at least see a notice: "Hey, this technician is viewing your screen." The cost one is about planning. With a clear price per endpoint, "I can build a product around it."

With both settled, his plan is simple: "I would try to push it across all my clients."

What he still wants, framed as polish: historical CPU, RAM and disk graphs to spot abnormal spikes; monitoring for Wi-Fi access points, switches, firewalls and websites; user activity logs next to agent logs, with the technician's name on remote-session logs; roles per organization; patch status and the policies that apply; compliance checkpoints, like marking a device as PCI DSS; suggested fixes for vulnerabilities; and technician takeover of Fae chats.

What he expects next: about 50 endpoints for sure once both conditions are met, and possibly another 1,500 across his clients.