Raul Moriarty

Background

As Communications Lead for Poker Bot AI — the umbrella organisation behind this site — I work at the edge of a fairly opaque part of the poker software world. The job is part product, part research communication, and part filter against the steady volume of marketing claims that pass for technical writing in this space. Before this role I worked alongside engineers building AI decision engines, with operators trying to defend their platforms against the same software, and with the end-users who simply want to know which of the things they read on a forum are actually true. The end-users are who this site is written for.

Most of the public material on club-app bots, hacks, and cheats falls into one of two failure modes. It is either marketing copy from sellers who want to present a generic solver as a system-level exploit, or it is forum chatter that lumps together a serious decision engine with deck-prediction snake oil and calls the whole field a scam. Both readings are wrong. The honest picture is that the engineering inside a working club-app bot is real and difficult, the platform-side detection is sophisticated but not magic, and the agent-and-club layer is the structural feature that makes the ClubGG environment different from a public operator.

Areas of focus

The threads I keep returning to in the club-app world:

Club-app bot architecture

Solver-anchored baselines from CFR-family solvers — PioSolver and GTO+ for heads-up and 6-max, MonkerSolver for multiway and PLO — compressed for runtime querying on mobile, paired with opponent models that converge fast under stable club identities. The interesting differences from public-operator bots are in the opponent-modelling step and in the UI automation layer.

The NSUS ecosystem — GGPoker and ClubGG

Shared corporate ownership and shared infrastructure produce a detection picture that has to be modelled across both products. The interesting research questions live at the cross-product boundary — signal sharing, abuse-graph propagation, transfer learning between population baselines.

The agent and club-owner enforcement layer

The structural feature that distinguishes club apps from public operators. A low-volume, high-context human classifier sitting on top of the platform's statistical stack. Most real-time bot identification at small and mid stakes runs through this layer, not through the NSUS network layer. Modelling it as an additional adversarial-classification stage is one of the open questions on this site.

Detection from the operator side

Behavioural fingerprinting, statistical play-pattern analysis, account-graph and showdown-correlation models, and the human review pipeline that signs everything off. Pure-GTO outputs are paradoxically easier to flag than noisier strong-human play because population variability is the baseline, and the cost matrix for false-positives is asymmetric.

Outside-channel collusion

The defining cheating vector on club apps. WhatsApp, WeChat, Telegram, Discord, KakaoTalk — channels the platform cannot observe directly. The defence is statistical and the operational challenge is setting per-club priors correctly without burning legitimate regulars.

Game theory in practice

Where the math says "stop." Some spots are solved well enough that further automation is rounding error; others — deep-stacked multiway turn play, ICM-heavy short MTT fields, Diamond Race multiplier dynamics — are still meaningfully open. Knowing the difference is part of taking the field seriously.

About this site

Three long-form notes (hacks, detection, FAQ) plus the home page cover what I think is worth saying publicly about the ClubGG ecosystem right now. Pages are revised when the field changes; the date at the top of each piece is the last revision, not the original publication.

The chat-link in the footer is the best way to reach me directly. Implementation questions, corrections, observations on cross-NSUS enforcement you have seen, references to academic work on adversarial classification or club-app detection — all welcome. The Poker Bot AI team reads everything. Product-pitch messages are auto-archived.