Ai BEAM benchmark - a fair look at where we stand on long-term memory We re-ran BEAM with our new context engineering and put the numbers next to the paper baseline. Honest wins, honest losses, and a clear roadmap.
