Run it on your own kit, then tell us what's wrong with it.

Who we're looking for

One shape of team, specifically — because that's the case LinkMesh was built for and the case we most need to get right — plus one variant of it:

• Your fleet's control plane can't live in a vendor cloud — either the network is genuinely isolated, or a policy names where control-plane metadata may reside. Banks, insurers, public sector, healthcare, critical infrastructure, industrial networks.
• You run enough collectors that managing them by hand hurts — roughly 25 and up, across more than one environment — your own, or your clients'.
• You have a real environment to point it at, not a slide about one. A lab or a staging fleet is fine; a hypothetical is not.
• Someone on your side can spend an hour or two with us and say what they actually think.
• Or you run fleets you do not own. You design, roll out and often keep running collector fleets across client estates — as a consultancy, an integrator or a managed-service provider. The sovereignty constraint is then your client's rather than yours, and what you need is something you can hand back intact when the engagement ends. That case is in scope, and the licence terms under What you get say so explicitly.

Who this isn't for

Where your telemetry goes isn't the dividing line. LinkMesh routes to cloud backends by design — Grafana, Splunk, Elastic, Azure, any OTLP destination — so a fleet that ships everything to your own cloud account is squarely in scope.

Where your control plane lives is the dividing line: the fleet inventory, the collector configuration, the credentials. If your collectors can reach a vendor-hosted control plane and you're content for that metadata to live there, the established cloud-hosted tools are the lower-friction choice and you should take them. We wrote that out in full in our Grafana Fleet Management comparison, and we meant it.

This programme isn't a discount scheme or a way to get a demo. If you just want to try the product, don't apply — download it instead. The Community tier covers 25 collectors with the full feature set and no time limit — free after a no-card registration in the OpenSight Customer Portal (5 without one) — and you never have to talk to us.

What you get

• An unlimited license for as long as we're working together, with no collector cap. This design-partner licence is a deliberate exception to LinkMesh's standard licence, which covers use within your own company only. It permits running LinkMesh for your own clients during the partnership; the licence stays with you and cannot be handed to a client. Reselling LinkMesh, or operating it as a service for third parties beyond this, requires a separate for-resale licence and a separate written agreement with us.
• A direct line to the people who build it. One address — info@opensight.ch — for anything at all: an install that won't go through, a config behaving oddly, a question about whether something is even possible. It reaches the engineers writing the code, not a ticket queue, and a person answers.
• Real influence on what gets built next. We're early enough that this is true rather than flattering.

What we ask in return

Your honest assessment — particularly the unflattering half. What was confusing, what was missing, what you'd have to explain to a security reviewer, and what would stop you rolling it out. An install that goes badly and gets described accurately is worth more to us than a polite one that goes well.

We're not asking for money, a public case study, a logo on our homepage, or a reference call. If you later want to do any of those, that's a separate conversation and the answer can be no.

What LinkMesh actually is today

So you can judge whether it's worth your hour before you write to us:

• It runs on your infrastructure. A single binary, a Docker image or a Kubernetes deployment, with embedded storage — no external database and no worker tier to stand up, and no calls out to a CDN at runtime. The control plane and every piece of fleet metadata stay inside your perimeter.
• Collectors authenticate with a durable Bearer token over TLS you terminate at your own ingress. Enrollment is an explicit onboarding step, not an open registration endpoint.
• Credentials live in a built-in vault and are rendered only at delivery, so secrets never land in stored configs and never get committed to Git.
• Pricing is per collector, not per gigabyte — so the bill doesn't move when your telemetry volume does.
• It's early access, and it has edges. The Trust page lists what we do and explicitly don't do, and the features page has the detail. Finding the edges is rather the point of this programme.