Mach9 Poker has closed. The app is off the store, the servers are off, and this page is what remains.
"It was time to let it go."
In June 2026 I went to the World Series of Poker. I ran into a lot of players, talked to a lot of them, and put the app in front of some of them at the table. Somewhere in those conversations the picture became clear.
Mach9 was built on one bet: that a fast tournament simulator and a field of 449 tuned AI opponents was something a human coach could not offer. Generative AI erased that bet from both sides. The coaches who already have the credibility and the audience now have the analysis too. The era of GenAI had almost certainly eliminated any chance of this product finding a paying audience.
And I had spent more than enough time trying. Several pivots, two platforms, a complete engine, a working app on a real iPad. Each step was reasonable. None of them answered the question that mattered, which is why anyone would need this one.
So this is a stop, not a pause and not another pivot. I am grateful to everyone who tested it, argued with it, and told me what was wrong with it. You made it better than it would have been.
David Cornelson
Founder, Mach9 Poker
The product is gone. The engineering was real.
A Rust tournament simulator: multi-table, dynamic balancing, side pots, bitwise hand evaluation. A full 450-player tournament ran in under two seconds.
A field of AI players with distinct personalities, from nits to maniacs, tuned against real VPIP, PFR, and aggression profiles until every one graded out.
A harness that decomposed which factors actually move decision value. It found that position is a modifier that depends on hand and board, not a standalone rule.
A SwiftUI learning ladder for iPad with commit-before-reveal practice and mastery gating, running the Rust engine natively on device.
The question was never whether the engine could be built. It was whether the engine would still be scarce when it shipped. It wasn't.
Machine-generated reasons for a play are often wrong, and a training tool cannot afford to be wrong exactly there.
The measurement work came last. It should have come first. It would have changed what we taught.
Changing how the app taught was never the same as re-establishing why anyone needed this one.
Thank you. The app has been removed from TestFlight, and all accounts, hand histories, and strategies have been permanently deleted from our servers.
The simulator and the AI field stand on their own and are being carried forward as a small browser experiment, separate from this product.
If you need something from us, email is still open.