"Vendor-agnostic" gets used loosely. Plenty of firms say it while quietly steering every engagement toward the one or two platforms they resell. We mean it literally, and the reason is structural: we have no platform of our own to push.
With nothing to sell you but the work, the recommendation is whatever fits. Sometimes that is the obvious answer. Often it is not.
The big three are the common case, not the limit
AWS, Google Cloud, and Azure cover most of what most companies need, and most of our work lives there. But "most" is not "all." We also work across Oracle Cloud and IBM Cloud, leaner providers like DigitalOcean and Hetzner where they fit the budget and the workload, edge platforms like Cloudflare, and private or hybrid setups on VMware, OpenStack, or Nutanix.
The point is not breadth for its own sake. It is that the right answer for a given workload is not always the platform a reseller happens to favor. A batch job that runs twice a day does not need the same home as a latency-sensitive service. A regulated dataset may need to stay on infrastructure you control.
Agnostic is also an operating habit
Being agnostic is not only about which logos we will work with. It shows up in how we standardize the things you control: consistent automation, one security posture, and one cost discipline across whatever platforms you end up running. That is what keeps multi-platform from turning into multi-headache.
Honest advice is easier to give when you are not selling the thing you are advising on. That is the whole idea.