← internals.quintace.ai
QuintAce · Media Partners

Long-form pieces challenging poker AI's assumptions.

In development for major media outlets. Each piece reframes how the field thinks about AI evaluation, what's actually winning real money against humans, and what it takes to learn poker beyond GTO solvers. Three angles, one underlying thesis: the only golden truth is real-game performance.

Active Drafts

v0.1 Draft

GTO Is a Toy. The Field Has Long Since Moved On: The State of Poker and Gameplay AI in 2026

The manifesto. The vendors are catching up to where QuintAce's DRL foundation model has been for years — GTO Wizard's April 2025 QRE launch is the latest sign. The academic descendants of Pluribus moved past Nash. The bots winning real money never used GTO. The methodology has long since moved on. Only the marketing hasn't caught up.

Track A1 Wired / Quanta / MIT Tech Review 4,000–5,000 words

In Planning — Manifesto Series Deep-Dives

Planning

Reading Solver Disagreement: A Methodology Deep-Dive

When commercial CFR tools disagree with each other and with DRL, the disagreement isn't a bug — it's the abstraction speaking. The 1755-flop alignment study; the ICM 0.28% vs 13% MAE finding; what disagreement actually signals. (Field-level companion to the Track B1 student piece.)

Track A2 Tech Review / IEEE Spectrum 2,000–3,000 words
Planning

Most Poker Games Have No Solver

Of 10 common poker formats, only one is essentially-solved. The architectural reason CFR can't catch up. PLO multiway, MTT-with-real-ICM, Squid — what the future of poker AI looks like when the toolchain stops at HU NLHE.

Track A3 Quanta / Tech Review 2,000–3,000 words
Planning

What the Russian Bots Know That Your Solver Doesn't

The only AI that consistently wins real money against humans uses exploitation, not GTO. Inside the bot wars: BFC, BEX, and what real poker AI looks like when you build it for the actual game.

Track A4 / Anti-cheat Bloomberg (Chellel) 2,500–3,500 words

In Planning — Standalone Angles

Planning

Why Studying GTO Won't Make You a Poker Player

Five years of GTO solver evangelism has produced a generation of players who can recite equilibrium frequencies and still lose money. What the gap between studying theory and making decisions teaches about learning anything difficult.

Education angle The Atlantic / New Yorker / HBR 2,500–3,500 words
Planning

The Bots That Beat Humans Don't Look Like the Ones in Science

Standalone investigative piece extending Bloomberg's BFC reporting. What anti-cheat engineers see that solver researchers miss. AceGuardian-led; pitches from inside the bot-detection war.

Anti-cheat / AceGuardian Bloomberg (Chellel) — primary 2,500–3,500 words
Planning

AI Still Needs You. AI Still Hasn't Beaten You.

Two open problems in modern gameplay AI — autonomy and outperformance. The best AI still needs human judgment in the loop (training, evaluation, edge cases), and still hasn't been demonstrated to outperform top human play in any published controlled setting. Coach-affirming, evidence-grounded.

Coach-affirming standalone Wired / HBR / Atlantic / poker community 3,000–4,000 words

In Planning — Student Series (Track B)

Practical, pedagogical companions to the manifesto. Same evidence base, gentler voice, written for poker students and coaches teaching with these tools. Distributed in poker community publications.

Draft v1

When QuintAce and GTO Wizard Disagree

Entry point. The student opens QuintAce, opens GTO Wizard, sees a divergence. This article walks through why disagreements happen, what they mean, and a 5-question diagnostic checklist for reading any solver disagreement intelligently. Anchored on postflop ICM as the worked example.

Track B1 TwoPlusTwo / Run It Once / GipsyTeam / QuintAce blog ~1,900 words
Planning

What's Actually Inside a Solver Solution?

The "let me show you the gears" piece. The tree designer is a person. The abstractions are choices. Vendor-published quotes do the heavy lifting. Pedagogical, not accusatory: this isn't a flaw — it's how the math works. You just need to know.

Track B2 Poker community + QuintAce blog 2,000–2,500 words
Planning

The Formats Your Solver Doesn't Really Cover

Format-by-format tour for students. Why does my solver feel less reliable in PLO / MTT / multiway? Multi-way pot dynamics solvers can't model; MTT-ICM snapshots vs real dynamic ICM; PLO tree explosion; Squid (no solver exists). Study guidance per format.

Track B3 Format-specific publications 2,000–2,500 words
Planning

Beyond the GTO Floor: Studying Real-Opponent Adjustments

Why pros tell students "GTO is the floor, not the ceiling." Population deviation patterns; how to study exploit lines; using QuintAce DRL exploit features as case study; the workflow that combines GTO foundation with exploit refinement.

Track B4 Coach + advanced student publications 2,000–2,500 words
Planning

How to Tell If a Poker AI Is Actually Good

The capstone. The four-rung evaluation ladder, taught as a student tool. When a tool says it's accurate, here's what that actually means. Apply to QuintAce honestly; demand the same disclosure from any tool you study with.

Track B5 Sophisticated student + coach publications 1,500–2,500 words

In Concept — Coach-Voice Series

Bylined first-person pieces by named pros. Edited by us, voiced by them. Each one extends the manifesto with primary-source authority.

Concept

What Pluribus Got Wrong (From Inside)

Bylined: Nick Petrangelo. First-person account of being a Pluribus participant + the methodology issues + the payment refusal context.

Track C1 Pending coach permission
Concept

The 7-and-Quarter BB Problem

Bylined: Uri Peleg. What he learned discovering DRL's action-set ceiling while writing the Squid invisible-ante piece.

Track C2 Foundation laid in existing article
Concept

Multi-Equilibria in 6-Max

Bylined: Dan Dvoress (proposed). Why 6-max NLHE has multiple GTOs and how that changes how he studies.

Track C3 Pending outreach
Concept

Live ICM and What Solvers Don't See

Bylined: Patrik Antonius (proposed). High-stakes live MTT dynamics that no current tool models.

Track C4 Pending outreach

Sequel / Future Concepts

Sequel

How Exploitable Is Your Solver, Actually?

A methodology paper replicating Lisý & Bowling 2017 for current commercial CFR tools (GTO Wizard, PIO, Monker post-2017). Cross-abstraction exploitability quantified for the first time. Academic publication track.

Sequel paper NeurIPS / AAAI / Science Q3 2026 target
Future Track

Industry / Operator Series

AceGuardian-branded series for poker rooms, operators, integrity bodies, investors. What AI means for real-money poker integrity, profitability, and the next decade of the industry.

AceGuardian Brand-strategy pending
Future Track

Pro / High-Stakes Series

For mid-to-high-stakes regs and pros. Population-aware play, game selection, and real-money winrate optimization beyond GTO. Sequenced after manifesto for poker-community amplification.

Future Poker community publications