Aigarth Is a Chess Game in Its Endgame. Here's the Board.

A different way to see where this project really stands. Not a roadmap, not a price chart - a game. One that started with a quiet opening twenty years ago and is now a few moves from mate. Every move below is a real, dated quote from the man playing it.

Why a chess game

A progress bar tells you how much. It doesn't tell you whether the position is winning. Chess does both. In a game, you can be at move 40 of 45 and already know it's over - the pieces are placed, the threats are unstoppable, the opponent is just delaying the inevitable. That is exactly where Aigarth is: deep in the endgame, with a position most people watching the scoreboard can't read.

The opening (2002-2022) - quiet, long, underestimated

Great games start slow. Pieces develop before anything dramatic happens, and to an outsider it looks like nothing.

Twenty years of developing pieces before the first capture. Then the board went live:

And the clock started on 13 April 2022 - real mining begins, the engine that powers every move from here. The opening is the part nobody claps for. It's also the part that decides whether you have a game at all.

The middlegame (2023-2025) - the pieces start falling

This is where a chess game gets decided - not by one brilliant move, but by a series of exchanges that quietly win material. Watch the verbs escalate. Looking for → found → we know how. Each one is a piece coming off the board in Aigarth's favor.

First exchange. Then a piece nobody expected:

Then the move that splits the game in half:

The midpoint. From here, the player stops developing and starts converting. And then, mid-2025, the two captures that decide the game - generalization and the universal training algorithm, both solved. In chess terms: the opponent just lost their queen and a rook. The position is now winning. CfB's own scoreboard:

Read that like a chess player would. "We know how to create AI" = the win is found. "We created none yet" = the mate isn't delivered. The hard part - seeing the win - is done. What's left is execution.

The endgame (2026) - converting a won position

Here's the thing about endgames: they look quiet, but they're the most precise phase of the game. One wall remains, and it isn't "can we win" - it's "can we win cleanly, without burning the house down." Energy.

The win is secured; now it's about the technique of finishing. And the technique is landing - the "constant dimensions" discovery, the anti-attractor mechanism, each one trimming the energy cost. Then the cleanest statement of a won endgame you'll ever read:

"Solved" plus "just a matter of execution" is what a grandmaster says when the board is won and only the moves remain. And the moves are happening right now: in June 2026 the score hit zero on the first tests, and CfB distributed seven executables for the coordinated final push - seven pawns, each one steps from promotion.

I opened one of those files myself. Inside: a trinary, sparse network - ~94% zeros, ~6% active connections. That sparsity is the energy efficiency, sitting in the bytes. You can see the winning technique on the board, not just hear about it.

The other player at the board - the scientists

Here's what makes this more than one person's claim. In chess, a won position is a fact - any strong player looking at the board agrees. Aigarth's position has been shown to other strong players, and they agreed.

The same paradigm Anna runs on has an academic twin: Neuraxon, formalized by David Vivancos and Dr. José Sánchez, working together - building the papers, the proofs, and the public case side by side. And the reviewers - people with no stake in the token - looked at the board:

20-22 May 2026, Berlin (ICMLT, IEEE): Neuraxon won Best Paper + Best Presentation, judged by independent scientists. Outsiders looked at the position and called it the best in the room.

27-30 July 2026, San Francisco (AGI-26): the Multi-Neuraxon paper accepted to the world's primary AGI conference - Vivancos and Sánchez noted it sits higher than the previous venues.

Plus Osaka (AMLDS, July) and ALIFE 2026 (August) - the winning streak continues across the field.

That's the difference between a player claiming a won position and a room of grandmasters confirming it. Marketing is free. Best Paper at an IEEE conference is not. The science is the second player nodding at the board and saying: yes, this is winning.

The board right now

Where the game stands, honestly:

Opening (idea, viability): done.

Middlegame (the neuron, the midpoint, generalization, training): done - the decisive captures are made.

Endgame (energy, practical proof): in progress, technique landing, first zeros on the board.

Mate: dated. 13 April 2027 - five years to the day from the first move.

If you put a number on it the way you'd assess a chess position: this is a won game in the endgame - roughly the last eighth of the moves, the won-position-conversion phase. Generalization and training are off the board. Energy is being converted right now. Seven pawns are one step from queening.

And the honest caveats, stated like a real player:

"Solved" means the win is found and being executed, not that mate is on the board today. The goal was always AI, not AGI - "the difference is the same as between a car and a BMW."

Zero on the tests proves the paradigm on the demonstration game - it's the won endgame, not the whole tournament.

13 April 2027 is the move clock. If mate isn't delivered by then, the position was misread. That date is the falsifier - which is exactly what makes this an honest assessment and not a promise.

The last move

It is easy to follow this game by the clock - the price, the daily noise. But chess is won on the board, not the clock. And the board is winning.

Two decisive captures already made. One conversion in progress, with the first zeros on the board. A row of independent grandmasters in Berlin and San Francisco confirming the position. And a move clock pointing at one date:

13 April 2027.

The opening was long and quiet. The middlegame won the material. The endgame is being converted, move by move, by our own hands on the keyboard.

It's not mate yet. But any player who can read the board can see it from here.

Watch the board, not the clock.

Sources: all quotes verbatim from CfB on Discord and X, dated inline and byte-verified against the project's own timeline. The chess framing is my own lens. Neuraxon results: Vivancos & Sánchez; ICMLT Berlin, AGI-26, AMLDS Osaka, ALIFE 2026. Anna.data structure observed directly. "Won endgame" describes the paradigm proven on a demonstration task, not AGI delivered; 13 April 2027 is the stated falsifier, not a guarantee.