Why a public download was never the right artifact
The domain says "download". The site doesn't host one. That's not coyness — it's the conclusion we came to after spending real time, in the early years, exploring what a public binary would look like for the kind of work we do. We didn't ship it. These are the reasons.
One binary, many rooms — the fit problem
The first reason is the simplest. A poker bot binary that runs against one operator's tables is the wrong binary for another's. The table size differs. The locale differs. The hand-history format differs. The detection vendor sitting in front of the client differs. The operator's tolerance for false positives on their own honest players differs. A public download forces every one of those decisions into a single artifact, and the artifact that results is either useless to most of the people who grab it, or unsafe — usually both. We don't see a way to ship one binary that's right for the kinds of rooms we actually work with.
Security: a public binary is a target, not a product
The second reason is what happens to a public binary after it's out. We've watched, from a distance, two competing teams discover that the moment a binary becomes downloadable, it stops being theirs. It gets repacked. It gets bundled with a side-loader that drains the client wallet. It gets reverse-engineered into a detection signature by the platforms it was meant to run against, and three months later the original team is patching defenses they didn't design.
None of those failure modes apply to a binary that only ever lands on a client machine we've already had a conversation with. We can sign it for that client. We can rotate the build identity. We can pull the previous artifact back if we have to. None of that is possible after a binary has been pulled to ten thousand machines we don't know about.
Legal exposure scales with the audience
The third reason is legal posture. Operator engagement is a contracted relationship with a counterparty in a known jurisdiction, with terms about what the binary does and doesn't do, and what the operator is expected to do with it. A public download is a relationship with the entire internet — including jurisdictions where the use case isn't on the same legal footing it is for our actual clients. Our legal contact's read, which we've never had a reason to argue with, is that you cannot draft terms-of-use for an artifact you've made available to everyone and expect those terms to protect anyone. We chose the smaller relationship on purpose.
Maintenance: who do you patch?
The fourth reason is operational. Our release cycle assumes we know exactly which clients are on which binary, because when a platform changes detection posture on a Wednesday, we need to know by Thursday morning which artifacts are exposed and which aren't. A public download breaks that lookup. If we cut a patch, we have no way to deliver it to the people who pulled the previous version off a page. We'd be shipping known-stale binaries into rooms we have no visibility into, which is the failure mode we built the whole release cycle to avoid.
What we do instead
We talk to operators directly. A first conversation is usually a thirty-minute call with the engineer on rotation — not a sales call, because we don't have a sales function. They'll tell you honestly whether your room mix is something we'd take on, or whether the right answer is somebody else. About a third of the conversations we have end with us pointing the operator elsewhere, which we'd rather do upfront than discover six weeks into a contract.
Engineer on rotation, not a sales inbox.
One contact. We answer release questions, scoping questions, and the "would you even take this on" question — without the pitch.
Reach the release engineer