200 realistic opponents. 33 data fields per action.
The richest low-stakes cash game data on earth.
A full poker client with persistent NPCs that tilt, bluff, slow-play, and make the same mistakes you see at real low-stakes tables. Play. Export everything. Feed it to AI.
See your game clearly.
Every NPC has a persistent name, face, and playing style. They remember how you play. They get tilted after bad beats. They bully when they're winning. They tighten up when they're scared. They make the same mistakes you see at real .25/.50/1.00 tables — not solver-perfect play, not random noise. Real human poker, modeled at a depth nobody else has attempted.
Tilt, steaming, confidence, fear — with intensity values that shift in real time. A bad beat at Table 1 affects their play at Table 3.
Every action has a reason: barrel continuation, street improvement value, low fear probe, sticky after investment. Not random weights — readable intent.
Snap calls, long tanks, hesitation patterns — each NPC has a timing profile with baseline speed, bluff speed bias, and pressure sensitivity.
Every hand is generated by the engine in real time. No two sessions are the same. Here are three moments from actual playtesting.
Giselle picks up pocket Kings and just calls preflop — classic TP slow-play. The flop comes low, she checks. The turn bricks, she checks again. Then on the river, after hero bets for value, she snap check-raises all-in. Tilt trigger: she'd been losing for an hour. The slow-play wasn't patience — it was a trap set by a frustrated player who finally woke up with a monster.
Tina calls preflop with K♠9♠ against hero's AK. Flop: A♠ J♦ A♣ — hero flops trip aces. Tina has nothing but a backdoor spade draw. Turn brings a spade, she tank-calls. River: fourth spade. Hero bets $18 for value. Tina tanks 6 seconds, then check-raise jams $87.20 all-in. A player who never does this — doing it at the exact right moment. $198 pot.
Tyler 3-bets to $11 with A♣K♣. Flop: J♣ Q♣ 10♣ — the royal flush. He raises to $44. Prashant Kumar, a loose-aggressive maniac, re-jams all-in with an open-ended straight draw. Tyler snap-calls. The nuts hold. $153 pot. A royal flush in a sim — proves the deck is real.
Every action from every player in every hand — with the full picture of why it happened.
{
"player": "Nadine Hewitt",
"actionType": "raise",
"amount": 42.50,
"street": "river",
"decisionTimeMs": 2847,
"timingTag": "tank",
"strengthBucket": "flush",
"confidenceTag": "high",
"decisionReason": "street_improvement_value",
"fearDiscount": 0.92,
"improvementTier": "major",
"betFracUsed": 0.85,
"callerResistanceProfile": "station_caller",
"momentumTags": ["aggression_inertia", "recent_success_confidence"],
"emotion": {
"state": "tilted",
"intensity": 0.73,
"trigger": "lost_big_pot"
}
}Why they did what they did. barrel_continuation, low_fear_probe, sticky_after_investment, pot_odds_rescue — readable intent on every action.
Tilt, confidence, steaming, fear — with intensity values and triggers. Know what each opponent was feeling when they made their move.
Decision time in milliseconds on every action. Snap calls, long tanks, hesitation patterns. Each NPC has their own timing fingerprint.
Fear discount per action — how scared an NPC is of the pot size, the board, and the opponent. Ranges from 0.0 (terrified) to 1.0 (fearless).
callerResistanceProfile — is the villain a station? A nit? A LAG? The engine knows, and sizes accordingly. Now you know too.
Pot, stacks, bet sizing, position, committed totals — before and after every action. Complete game state at every decision point.
Your data. Your tools. Your edge. We don't tell you how to use it — we give you the richest dataset in low-stakes poker and let you decide.
We're not competing with your favorite coaching tools. We're fueling them.
Every export writes four files directly to your machine. Structured for humans, machines, spreadsheets, and AI.
Human-readable hand histories. Full seating, HUD stats, emotional states, every action with pot/stack/timing/meta. Drag into ChatGPT and say "analyze my play."
Structured JSON with 33 fields per action. Decision reasons, fear modeling, tilt states, momentum tags, caller profiles. The raw engine intelligence.
Per-opponent scouting report. VPIP/PFR, timing profiles, action mix, emotional history, showdown frequency — computed from your session.
Same NPC data as a spreadsheet. Opens in Excel or Google Sheets. Sort by snap%, filter by archetype, find the most exploitable players.
Launch the app, click Open Table. You're seated with realistic opponents in seconds. Open up to four tables and grind.
Low-stakes cash games against 200 NPCs with names, faces, HUD stats, and real behavioral tendencies. It feels like a real client because it is one.
One click. Four files. Full hand histories, opponent profiles, timing data, emotional metadata — structured and ready for AI analysis.
Generate thousands of hands of the richest poker data available — without risking your bankroll.
Already subscribed?