G2

Noah H.

Senior IT Infrastructure & Client Solutions Manager

Information Technology and Services

Mid-Market (51-1000 emp.)

11/4/2026

"Reliable, Compliance-Friendly Cloud ML with Transparent Workspace Cost Tracking"

4/5

What do you like best about VESSL Cloud?

What stands out most to me about VESSL Cloud is how well it fits into the messy, real-world reality of managed IT services. We’re not a pure AI shop—we’re the guys who have to make sure a client’s cloud infrastructure doesn’t fall over while also keeping them compliant with PIPEDA. VESSL actually respects that balance.

Take a recent project with a mid-sized healthcare client in Ontario. They wanted to train some diagnostic imaging models, but their internal IT was a patchwork of on-prem servers and a half-finished Azure setup. We needed something that could abstract all that away without forcing us to rebuild their stack from zero. VESSL Cloud let us spin up isolated GPU workspaces with strict role-based access controls in about two days. For context, the last time we tried something similar with a competitor, we spent three weeks just fighting Kubernetes manifests and IAM policies. Here, the team was running experiments by day three, and I didn’t have a single support ticket about “can’t connect to the cluster” for the rest of the quarter.

The other thing I genuinely appreciate is the billing transparency. We manage IT budgets for retail and local finance firms where a surprise cloud bill is basically a fireable offense. VESSL’s granular cost tracking per workspace means I can show a client exactly what their data science team is burning through—no vague “compute units” that make my CFO squint. Last month, we caught a runaway training job on a Friday afternoon because the usage dashboard flagged it in real time. That alone saved us about $1,800 in weekend GPU charges.

It’s not flashy, and it doesn’t try to be. It just works, scales down when we need it to, and doesn’t make my team babysit infrastructure. For a 51–200 person MSP juggling healthcare compliance and retail uptime, that’s exactly what I want from a platform. Review collected by and hosted on G2.com.

What do you dislike about VESSL Cloud?

If I have to nitpick, my biggest frustration with VESSL Cloud is the documentation and the overall onboarding ramp for non-ML specialists. We are an IT managed services provider, not a dedicated data science lab. My team handles enterprise cloud infrastructure, network security, and SaaS rollouts. We don't live and breathe PyTorch or CUDA daily.

A prime example happened just last month during a rollout for a mid-market retail client in BC. We were setting up a custom IAM policy to restrict data access for a third-party contractor, which is standard practice for our PIPEDA compliance workflows. I wanted to find the exact JSON parameters for a custom role, but the VESSL documentation assumed I already knew their exact internal terminology. I had to dig through community forums and eventually file a support ticket to get a straightforward answer. With AWS or Azure, I can usually find a clear Terraform template or a precise API reference in minutes. With VESSL, I often feel like I'm reverse-engineering the intended workflow.

Additionally, the native alerting could be much tighter for our use case. We had a scenario where a client's data engineering team left a heavy GPU node running over a long weekend. While the cost dashboard is great for post-mortem analysis, I really wish we could set up more granular, proactive budget threshold alerts. I got a notification, but by then, the damage to our monthly margin was already done. I need the system to automatically throttle or pause non-critical instances the second they hit a pre-set spend limit, rather than just sending an email that I might not see until Monday morning.

Also, the UI/UX can occasionally feel a bit clunky when you're jumping between multiple client workspaces. It’s a minor gripe, but when I’m managing three or four different environments for various clients, the navigation feels less intuitive than the major hyperscalers. That said, these are operational growing pains. The core platform is solid, but for an MSP trying to enforce strict compliance and predictable billing, the lack of deeper, enterprise-grade documentation and automated cost guardrails is definitely the friction point we bump into most. Review collected by and hosted on G2.com.

What problems is VESSL Cloud solving and how is that benefiting you?

The core problem VESSL Cloud solves for us is the gap between a client's messy, homegrown data science setup and the compliance-grade infrastructure we're contractually obligated to deliver. A lot of our mid-market clients in Ontario and BC—retail chains, local finance firms, a couple of healthcare clinics—have data science teams that were basically self-hosting. Someone's got a GPU box under a desk, models get trained in a notebook on a laptop, and nobody can tell me who touched what data when. That's a PIPEDA audit waiting to happen. VESSL gives us a single, auditable control plane: isolated workspaces, role-based access, and resource provisioning that I can defend to a client's security committee. That's the problem—ungoverned experimentation—and the benefit is that we can now say "yes" to AI workloads we previously had to decline.

Concrete example: a BC retail client came to us wanting to fine-tune a demand-forecasting model on three years of transaction data. Their previous attempt stalled because their single on-prem GPU was constantly tied up, and their team had no way to queue or partition workloads. We provisioned two isolated VESSL workspaces—one for experimentation, one for staging—with scoped access for the client's team and full audit logging for us. Training time dropped from days of serialized waiting to parallel jobs finishing in a few hours, and more importantly, when their compliance lead asked "who accessed the customer dataset last Thursday," I pulled the log in about ten minutes.

The second problem it solves is cost opacity. Our clients don't have dedicated FinOps people. VESSL's per-workspace usage tracking means I can walk into a monthly review and show exactly which experiments consumed which GPU hours, instead of guessing at a blended cloud invoice. That's benefited us directly in renewal conversations—we justify our managed services fee with actual resource attribution rather than hand-waving.

The honest caveat: none of this is magic. You still need someone who understands IAM, network segmentation, and data residency to configure it properly. But that's our job as the MSP, and VESSL at least gives us the primitives to do it without hand-rolling Kubernetes for every client. Review collected by and hosted on G2.com.