Call +21% facing a flop raise. Check-raises 28% ± 6 against a 9% baseline, mostly draws and air.
Poker RTA
that knows
the player.
Real-time assistance that reads the table, solves the spot and adjusts to the player in front of you.GTO is the foundation.The player profile is the edge.
No Installation. In Browser. Runs on our 40 TB Server.
01 / THE PLATFORM
Read the table.
Model the opponent.
One system for pixel-level table recognition, solver-led decisions and the statistical fingerprint that makes every opponent different.
Every decision
starts with a read.
Your cards. The board. Position, stacks, the pot and the price to continue, read from the pixels of the shared window several times a second. No OCR lag, no manual input, amounts normalised to big blinds.
- Dedicated detectors for GGPoker, ClubWPT Gold, ClubGG and PokerNow
- Street, dealer button, action order and who is still in the hand
- Card votes across frames, so a mid-animation flip never misreads
TAG · 140 hands · 78%Pot 39.5 bb · TiltedTom to actButton on RiverRat_77Certain fold marked before the action arrivesEquilibrium first.
Deviate with a reason.
Every spot starts from the game-theory-optimal line: position, effective stacks, sizing and the ranges each player arrives with. You see the mixed strategy as frequencies, not a single guess, so you know exactly what a 55/45 spot is.
- Preflop charts by format and stack depth, push/fold when short
- Postflop frequencies with equity, alpha and minimum defence built in
- Certain folds shown before the action even reaches you
Top pair, weak kicker: a bluff-catcher. Equity 31% against 29% needed. α = 0.40, so the call is indifferent at this sizing.
The solver sets the line.
The opponent model bends it.
Every opponent tendency is a Beta-Binomial posterior: a prior from the population, updated with every hand you have seen from them, with the uncertainty carried along. Once the evidence beats the noise, the exploit is sized by how far they sit from the theory: folds beyond alpha, bluffs beyond what a bluff-catcher needs.
- Archetype classification: TAG, LAG, nit, calling station, maniac, glass cannon
- Exploit shifts capped and damped by sample size, never a hunch
- GPT-5 turns the numbers and your notes into a line you can act on
Bluff-catcher at indifference. 12 hands on Northstar is not evidence yet.
Your note "bets big with air" agreesNew session.
Same player memory.
Every hand you play is logged under your login: showdowns, sizings, positions, streets. Profiles follow the player across sessions and across their screen names, and your own notes change how the model reasons about them.
- Searchable player log with editable names and aliases
- Notes that feed straight into the exploit reasoning
- Confidence bars per player, sized by the hand sample
Click a player to open the model read, the confidence and the notes. Notes change the coach's advice against that player from the next decision on.
Loose-passive: too many hands, too few raises, and a river that never folds. Value-bet thinner, bluff less, call his big bets wider.
- Calls 58% of hands preflop, raises only 12%: a limp-call range full of dominated hands
- Folds to a river bet 22% of the time; the bettor needs 40% folds at 2/3 pot, so bluffing him is a losing play
- Bluffs 46% of river bets at showdown, well over the 29% a bluff-catcher needs
Your notes, read by the coach: "calling station", "bluffs a lot", "never bluffs", "nit", "folds too much" each shift the advice against this player.
Tight-passive: opens only premiums, folds to pressure and rarely bluffs a river. Steal his blinds, c-bet bluff more, fold to his big raises.
- Opens 11% of hands and folds to 3-bets 71%: 3-bet bluffing his opens shows an immediate profit
- Folds to c-bets 70%; a third-pot bet needs 25% folds, so any two cards can bet the flop
- Raises the river 4% of the time, almost always with the nuts: fold the second-best hand
Your notes, read by the coach: "calling station", "bluffs a lot", "never bluffs", "nit", "folds too much" each shift the advice against this player.
Solid regular. Plays close to the pool on every axis; the edge is small and comes from the 3-bet spot.
- Opens 20% and folds to 3-bets 61%, above the 57% a 3x raise needs: 3-bet bluff her opens
- C-bets 64% and gives up on 48% of turns when called: float the flop, bet the turn
- Reaches showdown 25% with a value-heavy range: her river bets are rarely bluffs
Your notes, read by the coach: "calling station", "bluffs a lot", "never bluffs", "nit", "folds too much" each shift the advice against this player.
Aggressive 3-bettor who check-raises flops with draws and air far more than the pool. Gives up on turns when called. Folds to flop re-raises more than he should.
- Check-raises flops 28% against a 9% baseline, mostly draws and air: second pair is ahead of most of it
- 3-bets 14% of hands and folds to 4-bets 48%: 4-bet bluffing his 3-bets is profitable
- Calls flop re-raises 70%: re-raising as a bluff loses, re-raise for value only
Your notes, read by the coach: "calling station", "bluffs a lot", "never bluffs", "nit", "folds too much" each shift the advice against this player.
Low evidence. Loose and passive so far: value bet thinner once the sample grows, bluff less now.
- Limps 46% of hands and raises 6%: a limp-heavy range of weak broadways and suited junk
- Folds to c-bets 25%, under the 33% a half-pot bet needs: c-bet bluffs lose for now
- 23 hands: the model leans on the 9-max pool until the sample grows
Your notes, read by the coach: "calling station", "bluffs a lot", "never bluffs", "nit", "folds too much" each shift the advice against this player.
Call +19% on river bluff-catches. Bluffs 46% ± 9 at showdown, 17 points over indifference.
3-bet bluff +8% against her opens. Folds to 3-bets 61%, above alpha for a 3x raise.
C-bet bluff +17%. Folds to c-bets 70%, far past alpha for a third-pot bet.
Low evidence. Value bet thinner once the sample grows. Calls everything so far.
No exploit armed. Nine hands is not evidence. Pure GTO until it is.
02 / THE MODEL
GTO solver. Bayesian opponent model.
Then the exploit.
Three layers, each one only allowed to move the line as far as the evidence supports.
The equilibrium line
For a bet of s pot, the opponent must fold more than α for a bluff to print; you must defend at least the MDF to deny it. The solver output is the mixed strategy at indifference for your exact hand, position and sizing.
The opponent posterior
Each tendency is a Beta-Binomial estimate: k observed events in n chances, shrunk toward the population rate b with a pseudo-count m. The standard deviation travels with it, and an exploit is only armed when the z-score clears the noise.
The sized deviation
The line moves in proportion to how far the opponent sits from the indifference point p*, capped, and damped by the sample. GPT-5 then reasons over the numbers, the archetype and your notes to explain the play in plain words.
03 / HAND HISTORIES
Arrive already
knowing the regulars.
Upload the hand histories you already have and every hand is added to the players' memory before you sit down, so the exploits are armed from the first hand.
Every hand counts,
even the ones before today.
Parsed on our server with no AI calls: showdowns, sizings, positions and streets all feed the same posteriors as live play. Every live hand is saved the same way, with the advice you were given, so a session can be reviewed decision by decision.
- PokerNow CSV exports and PokerStars or GGPoker hand histories
- Holdem Manager, PokerTracker and Hand2Note exports, from any room
- Telegram room hand recaps, in Chinese or English
04 / TABLE SIZES
Heads-up,
6-max or 9-max.
Up to four tables at once.
Built around no-limit hold'em cash games; tournaments are coming soon.
The same coach, a different table dynamic, chosen with one click.
Multi-tabling: share each table's window, and every table gets its own reader, its own advice and its own player reads, side by side on one screen.
Every hand is a battle of ranges. Wide opens, wide defends: the coach waits for the actual spot rather than pre-judging a hand, because almost everything plays.
Push/fold charts take over below 15 big blinds effective. Amounts are normalised to big blinds whatever the site shows.
05 / THE SITES
Works on any poker platform.
Dedicated readers for GGPoker, ClubWPT Gold, ClubGG, PokerNow, CoinPoker and PokerStars; any other platform on request.
Each site animates, deals and sweeps chips differently, so each one has its own detector, its own timing rules and its own test fixtures. Your platform not here? Message us on Telegram: we set up our AI to learn your table and fit a reader for it.
GGPoker
Cash and Rush & Cash tables, any table theme, on GGPoker or Natural8. Desktop client or browser.
ClubWPT Gold
Eight-seat tables in the browser, in chips or big blinds: cards, stacks, bets and the pot read on your device. New, built from players' tables.
ClubGG
Private club cash games. Built for the ClubGG table layout.
PokerNow
Home games and private clubs in the browser. Chips or big blinds, any number of seats.
CoinPoker
Cash tables in the CoinPoker client: cards, stacks, bets and the button read on your device. New, built from players' tables.
Your platform not listed?
Pick Others in the app, or message @GTOMAXAI with the name of the room: we set up our AI to learn your table and fit a dedicated reader for it, usually within days.
06 / HOW IT WORKS
Poker RTA in your browser.
No download, no HUD to install.
No desktop client, no installer, no API key of your own.
Open the app, sign in, share the table.
Sign in
Use the username and password you got on Telegram. Your settings, hands and players load with it, from any computer.
Share the table
Pick your site and table size, then share the poker window with the page. The coach pops out as a small always-on-top window.
Play
Every decision shows the GTO line and the exploit line with the reasons: equity, pot odds, the opponent's posteriors, your notes.
07 / PRICING
Two plans.
One with the player model.
Solver gives you the GTO line. AI Edge adds the opponent model, the exploit line and the reasoning. Free 30-minute trial on the spot. AI Edge by card, weekly or monthly, live the moment you pay; Solver and crypto payments on Telegram.
Solver
Real-time table recognition.
The GTO line on every decision.
- Live pixel reads on GGPoker, ClubWPT Gold, ClubGG and PokerNow
- Equilibrium frequencies for your exact spot, preflop to river
- Equity, pot odds, alpha and minimum defence on every line
- Certain folds shown before the action reaches you
- Heads-up, 6-max and 9-max cash · up to 4 tables at once · browser-based, no install
AI Edge
GTO at the core.
The player profile is the edge.
Card payment through Stripe, renews weekly, cancel any time. Prefer crypto? Message us on Telegram.
EVERYTHING IN SOLVER, PLUS- Bayesian exploitation modelling: a Beta-Binomial posterior for every tendency of every opponent, updated hand by hand
- Archetype classification with confidence: TAG, LAG, nit, calling station, maniac, glass cannon
- The exploit line, trained on over 4 million hands: deviations sized by how far the player sits from alpha and indifference, damped by sample
- GPT-5 reasoning that turns the numbers and your notes into the play, in words
- Player memory: log, profiles, aliases, notes and hand history under your login, across sessions
- Confidence bars per player and per read, so you know when the model is sure
AI Edge Monthly
Everything in AI Edge.
One payment a month.
Card payment through Stripe, renews monthly, cancel any time. Prefer crypto? Message us on Telegram.
EVERYTHING IN SOLVER, PLUS- Bayesian exploitation modelling: a Beta-Binomial posterior for every tendency of every opponent, updated hand by hand
- Archetype classification with confidence: TAG, LAG, nit, calling station, maniac, glass cannon
- The exploit line, trained on over 4 million hands: deviations sized by how far the player sits from alpha and indifference, damped by sample
- GPT-5 reasoning that turns the numbers and your notes into the play, in words
- Player memory: log, profiles, aliases, notes and hand history under your login, across sessions
- Confidence bars per player and per read, so you know when the model is sure
Start a free 30-minute trial of AI Edge before you decide: your email and a password, no card, no chat. Pay by card with the email of your account and the week or month lands on it at once; a crypto payment is confirmed on chain. Either way your account is extended from its current end date, so paying early loses nothing.
Cash games today; tournaments are coming soon. Access is personal. Read the terms of use before you play.
08 / WHAT YOU GET
Real-time assistance,
and what sits on top of it.
An RTA poker tool shows solver output while you play. GTO MAX AI does that, and models the player you are facing.
Reads the table live
Cards, board, stacks, bets, pot and the button are read from the shared window several times a second. The GTO line and the exploit line appear the moment the action is on you.
A profile for every opponent
Type, stats, showdown tendencies and your own notes, built from every hand you play against them. HUD-style reads on GGPoker, ClubWPT Gold, ClubGG and PokerNow.
Bayesian opponent model
Each tendency is a Beta-Binomial estimate with its own confidence. An exploit is sized by how far the player sits from alpha and indifference, and damped by the sample size.
Up to 4 tables at once
Share each table's window and every table gets its own reader, advice and player reads, laid out side by side on one screen. Browser-based: open app.gtomax.ai in Chrome or Edge, sign in, share. Nothing is installed.
09 / FAQ
Questions
people ask us.
Anything else: message @GTOMAXAI on Telegram.
What is poker RTA, and is this RTA?
Real-time assistance is solver-based advice while a hand is being played. GTO MAX AI shows the GTO frequencies for your exact spot and an exploit line adjusted to the opponent in front of you, live. Whether you may use it depends on where you play: most online sites forbid it, home games and private clubs decide for themselves. Read the fair-use terms in the app before you start. The full picture, from manual vs automated RTA to how sites detect it and how the tools compare, is in our poker RTA guide.
Is this a bot?
No. It reads and advises; it never clicks. Every action and every wager is yours. It does not place bets or play hands for you.
Can I multi-table?
Yes, up to four tables at once. Pick the number of tables, share each table's window, and each one gets its own reader, its own advice panel, its own player reads and its own floating coach, arranged side by side on one screen. The table where it is your turn lights up.
Which sites does it work on?
We support all platforms. GGPoker, ClubWPT Gold, ClubGG, CoinPoker, AAPoker, PokerNow and PokerStars each have their own pixel-level reader, kept separate so a change for one site never affects another. If your site is not listed in the app, contact us on Telegram at @GTOMAXAI and we will send you the instructions for setting it up with our vision reader.
What does "Bayesian" mean here?
Every tendency, such as how often a player folds to a river bet, is estimated as a posterior: a population prior updated with the hands the coach has seen from that player, with the uncertainty kept. An exploit is only applied once the evidence clears the noise, and its size is damped by the sample. GPT-5 then reasons over those numbers and your notes.
Do I need to install anything?
No. Open app.gtomax.ai in Chrome or Edge on Windows, macOS or Linux, sign in, share the poker window. The coach pops out as a floating window that stays on top of the table.
How do I get access and pay?
Open app.gtomax.ai, press Start free trial and sign up with your email and a password: 30 minutes of AI Edge, on the spot. For AI Edge, subscribe by card with the email of your account, weekly at $179 or monthly at $599 (22% less): access is live the moment the payment goes through, renews each week or month and can be cancelled any time. Solver, $79 a week, and crypto payments are arranged with @GTOMAXAI on Telegram; renewals add to the current end date.
Is this a HUD for GGPoker or Natural8?
Not a HUD in the classic sense: nothing is installed and nothing reads the client's memory or hand-history files. GTO MAX AI reads the shared window as pixels and builds its own player log, so you get HUD-style stats and reads on GGPoker, Natural8, ClubGG and PokerNow, where conventional HUDs do not run. Whether you may use it is between you and the site's rules.
Is it an alternative to GTO Wizard or a solver?
They complement each other. GTO Wizard and solvers are for study away from the table. GTO MAX AI is for the decision itself: the equilibrium line for the live spot, then the exploit line from the opponent model, then the reasoning. Use a study tool to learn the theory and GTO MAX AI to apply it against the player in front of you.
What is the best poker RTA?
Judge an RTA on three things: does it read the table itself, does it know the opponent, and does it explain the line. GTO MAX AI is the only real-time poker assistant we know of that does all three, live, on GGPoker, ClubWPT Gold, ClubGG and PokerNow, with a Bayesian opponent model and GPT-5 reasoning on top of the solver.
Cash games or tournaments?
Cash games, heads-up to 9-max, today. The push/fold charts already take over below 15 big blinds, and full tournament support with ICM is coming soon.
Where is my data?
Your hand history, player profiles, notes and settings are kept under your login on our server and shown to nobody else. Frames of the shared window are processed on our server to read the table and are not kept.
GTO MAX AI · CASH GAMES · GGPOKER, NATURAL8, CLUBGG, POKERNOW, COINPOKER, POKERSTARS, ANY PLATFORM
KNOW
THE PLAYER.
Get a 30-minute trial 30-minute trial on Telegram.
