Raul Moriarty

Background

As Communications Lead at Poker Bot AI — the umbrella organisation that runs this site — I've worked on the edge between poker software engineering and the people who play with or against it. The work covers AI decision engines on the real-money side of poker, operator-grade detection systems, and (the topic of this site) the much smaller and structurally different question of what bots and cheats look like on a free social app like Poker Now. Over the past fifteen years I've spent enough time with each of those audiences — engineers, operators, end-users — to know which questions are the interesting ones and which are recycled marketing claims.

Most public writing on "poker bots," "hacks," and "cheating" lands in one of two equally unhelpful camps. The marketing camp treats any solver-anchored AI as a "server exploit," because the phrase sells better. The forum camp treats every form of automated decision support as equivalent to deck-prediction snake oil. Both framings prevent readers from understanding what real poker software actually does, where the engineering problems are, and what an operator like GGPoker (or a free app like Poker Now) is and is not capable of doing on the detection side. This site exists to push past both.

Areas of focus

The threads I keep returning to across the poker app landscape:

Modern poker software architecture

Solver-anchored decision engines (PioSolver, GTO+, MonkerSolver, the Pluribus and DeepStack research lineage), compressed for run-time querying, paired with opponent models that converge within a session rather than over years of HUD data. The interesting engineering is in the compression and the online updates, not in the headline algorithms.

The poker app ecosystem

Real-money operators (GGPoker, PokerStars, partypoker, WPT Global) compared with free social apps (Poker Now, Pokerrrr2, similar). The shape of the bot and cheat market in each segment is a direct function of whether the platform has a rake stream — which is why the segments look so different from one another.

Detection from the operator side

The four-layer model that real-money operators run (behavioural fingerprinting, statistical play-pattern analysis, anti-collusion graph models, human review) and the reason free apps like Poker Now intentionally do not. Anti-detection treated as adversarial classification with an asymmetric cost matrix, not as a checklist.

Business and product

Fifteen years of building software gives a useful filter on which claims from poker AI marketing have any basis in reality. Most "Poker Now hack" listings, and a meaningful chunk of "poker bot" listings on Telegram more broadly, are sales copy with no engineering behind them. Saying that clearly has been more useful to readers than yet another neutral product roundup.

Game theory in practice

Where the math says "stop." Heads-up NLH at 100bb deep is solved well enough that further automation is rounding error. Deep-stacked multiway turns and ICM-heavy MTT endgames are still meaningfully open. Knowing the difference is part of taking the field seriously.

About this site

Three long-form notes (hacks, cheating and collusion, FAQ) plus the homepage cover what I think is worth saying publicly about the Poker Now 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.

There is a chat link for developers, home-game hosts, and curious players. If you have a question about something covered here, a counter-argument to one of the claims, or you are working on the real-money side of poker AI and want to compare notes, the chat is the right starting place. The Poker Bot AI team reads everything that lands there.