A native 6-max, 100BB training table where five GTO opponents play every hand through the same Pokerai API available to developers. Counterplay records every Hero decision, grades it against returned frequencies, groups recurring deviations into leak reports, and keeps the underlying hands ready for review or optional grounded coaching.
270lines — the whole API client
0cached strategy
10API endpoints in use
Updated · Maintained by Pokerai API
Counterplay / iOSreal product interface
01 / architecture
A poker engine owns the hand. Pokerai API owns the strategy.
Counterplay separates game mechanics, session state, and strategic truth. PokerKit advances the six-handed table; the trainer service preserves the ordered action line, effective stacks, ranges, and review context; Pokerai API supplies presolved lookups or real-time solver output. The same response can drive opponent play, Hero grading, range visualization, and later coaching without embedding private charts in the app.
01 / experience
iOS app
SwiftUI interface with PokerKit game mechanics and decision capture.
02 / state
Trainer backend
FastAPI holds the current hand, action line, ranges, and review context.
03 / strategy
pokerai.bet
Presolved preflop and flop strategy plus queued turn and river solving.
02 / product loop
Train at the table. Find the pattern. Reopen the hand.
The product does more than expose one solver response. It turns individual decisions into an analysis report, then keeps the underlying hands available for review and follow-up study.
Analysis report
See repeated deviations
Group similar decisions, show sample counts and average deviation, then rank repeated preflop and postflop patterns by severity.
Hand history
Return to the exact spot
Search by hand, position, or cards; inspect the result and net BB; then reopen the exact decision sequence behind a report.
Discuss and lookup
Study a spot without playing a full hand
Choose quick presolved lookup or an exact live solve, define the pot and positions, and carry the result into a focused discussion.
Trainer preferences
Control language, identity, and coaching model
Keep the training surface focused while letting the player choose locale, table identity, and the optional model used for explanations.
03 / endpoint ledger
Every street maps to a documented API surface.
Traced from the real client (method → endpoint → where it's used):
The client does not embed charts or solver output. It shapes the current spot, sends an authenticated request, and returns normalized strategy frequencies to the trainer.
01One shared authentication header.
02Typed methods named after public endpoints.
03One timeout and retry path for every street.
04No private strategy cache in the application.
app/gto/client.pythin client excerpt
class GtoClient:
def __init__(self, settings):
self._headers = {"Authorization": f"Bearer {settings.gto_api_key}"}
async def preflop(self, hole_cards, hero_pos, preflop_actions):
body = {"hole_cards": hole_cards,
"positions": {"hero": hero_pos},
"preflop_actions": preflop_actions}
return _strategy_from(await self._post("/v1/gto/preflop", body))
async def flop_node(self, node_token, hole_cards=None):
body = {"node": node_token}
if hole_cards: body["hole_cards"] = hole_cards
return _strategy_from(await self._post("/v1/gto/flop/node", body))
# … + solver_schedule / solver_tree / solver_node, all one-liners over _post()
05 / decision loop
From a dealt hand to grounded feedback.
01 / Preflop
Read the line
Send cards, position, and ordered actions; use returned frequencies for play and grading.
02 / Flop
Walk the tree
Load the presolved tree, resolve the active node, and project ranges after each action.
03 / Turn + river
Queue a solve
Hand ranges to the real-time solver, poll the job, then read the resulting strategy node.
04 / Review
Grade and explain
Measure the chosen action by its own GTO frequency and pass structured context to the optional coach.
Scope and provenance
Counterplay is a Pokerai API product integration maintained by Pokerai API. It is not presented as a third-party customer deployment. This technical record documents the iOS trainer workflow and the public API endpoints used by that workflow. The current product targets six-handed 100BB cash-game training, randomizes Hero's seat, uses five strategy-driven opponents, and reduces multiway postflop spots to a heads-up approximation. Coaching is optional. Counterplay does not enable real-time assistance at real-money tables. This page makes no claim about customer adoption, downloads, benchmark performance, commercial partnerships, or external endorsements.
Next case / 02
MelaGTO
See the same strategy layer become a native macOS workspace for range study, custom postflop solving, imported hand review, and training.
