Raul Moriarty

Background

I work as Communications Lead for Poker Bot AI, the umbrella organisation behind this site, sitting at the boundary between the engineers who build AI decision engines and the players and researchers trying to understand what those engines actually do. Fifteen-plus years across software engineering, business development, and online poker have given me a useful filter for the gap between what gets written about "poker bots" and what the systems actually are.

Most public writing on ACR Poker bots, hacks, and cheating fits into one of two failure modes. The first is marketing copy that claims solver-anchored decision software is a "server exploit," conflating engineering with magic. The second is forum reaction that treats any automated decision support as identical to deck-prediction snake oil, conflating engineering with fraud. Both frames prevent the reader from understanding what real poker software does, where the genuine engineering problems sit, and what an operator like WPN actually runs on its detection side.

Areas of focus

The threads I keep returning to in poker software:

Modern poker software architecture

Solver-anchored baselines built from CFR variants (PioSolver and GTO+ for heads-up and 6-max, MonkerSolver for multiway and PLO), compressed for real-time querying, combined with opponent models that converge within a session rather than over years of HUD accumulation. The compilation-to-deployable-engine pipeline is where most production engineering effort actually goes.

The ACR Poker and WPN ecosystem

Architectural features specific to this operator: fixed screen names (which make HUDs work for both player and operator), the four-skin player pool joining ACR with BlackChip Poker, TruePoker and YaPoker, the Curaçao licence and crypto cashout path, the post-Black-Friday US-facing position. WPN's public enforcement history — most prominently the 2015 KhanZ bot-ring takedown — is the most informative bust record of any major operator and structures how I write about detection.

Detection from the operator side

The four-layer model (hand-history forensics; behavioural fingerprinting; cross-skin account graphs; human review) and where naive software is caught. This is the area that benefits most from honest explanation — not as a "how to avoid bans" checklist but as an adversarial-classification problem with an asymmetric cost matrix and a non-stationary decision boundary.

Business and product

Fifteen years building and shipping software gives me a useful filter on how much of the "poker AI" claim landscape has engineering behind it and how much is sales copy. Most of what is sold as an "ACR Poker hack" falls into the sales-copy category, and stating that plainly has been more useful to readers than yet another neutral-tone product review.

Game theory in practice

Where the math says "this is solved" versus where the math says "this is still open." Heads-up no-limit and 6-max NLH at 100bb are solved well enough that further automation is rounding error against strong regulars. Deep-stack multiway turn play, ICM-heavy MTT endgames, and the PLO multiway frontier are meaningfully open. Knowing the difference is part of taking the field seriously.

About this site

Three long-form notes (hacks, detection, FAQ) plus the homepage cover what I think is worth saying publicly about the ACR Poker and WPN ecosystem right now. Pages are revised when the field changes; the dates at the top are last-revision, not original-publication.

The chat link at the bottom of every page is read by the team. Questions about anything covered here, corrections, or data we have not seen are all welcome. Sales messages auto-archive.