PRAXIST System Turns AI Experiments into Lasting Knowledge
The PRAXIST paper describes agents that store typed findings from each attempt for later use.
Quotes from researchers highlight the PRAXIST paper's approach of turning each attempt into typed evidence that subsequent agent generations can inherit, validate, recombine, or discard. One post notes its application to a tokamak magnetic control case study in fusion research. Another cites MLE-bench results of 60 medals including 49 gold at $3,054 total spend versus a baseline of 55 medals and 34 gold at $38,370. Posts contrast this with systems that discard intermediate results and link the method to Assembly Theory.
Complex engineering solutions are rarely the result of a single breakthrough. They are assembled from discoveries: a mechanism uncovered in one experiment, a design principle validated in another, or an interaction that later becomes useful in a completely different direction.…
Combined views
9 posts, first seen 2d ago
PRAXIST System Turns AI Experiments into Lasting Knowledge
The PRAXIST paper describes agents that store typed findings from each attempt for later use.
Quotes from researchers highlight the PRAXIST paper's approach of turning each attempt into typed evidence that subsequent agent generations can inherit, validate, recombine, or discard. One post notes its application to a tokamak magnetic control case study in fusion research. Another cites MLE-bench results of 60 medals including 49 gold at $3,054 total spend versus a baseline of 55 medals and 34 gold at $38,370. Posts contrast this with systems that discard intermediate results and link the method to Assembly Theory.
Complex engineering solutions are rarely the result of a single breakthrough. They are assembled from discoveries: a mechanism uncovered in one experiment, a design principle validated in another, or an interaction that later becomes useful in a completely different direction.…
Reactions
EV 
Evelyn@Evelynb7a1@Sapient_Int Great perspective on innovation—complex breakthroughs are often built through continuous discovery and connecting ideas across fields
CR Cryptoartmaster@VitalijMatros@Adidotdev treating failures as data instead of garbage is how agents actually scale intelligence.
RM Rahin Mazumder@boka_shoka@neil_xbt This is a much smarter way to evaluate autonomous research
|S | Shux |@ShuxBroh@alannnfx Banger tech