Llama Brain
A llama with its brain in a jar. 66 spiking neurons, wired to real Solana pools. It reads each token with its eyes, ears, nose and tongue, then decides: swallow, spit, yawn or keep chewing. Four minutes later it checks the chart and learns if it was right.
Field notes
Every verdict goes in the notebook with the numbers that made the neurons fire. Four minutes later the llama goes back to the chart and writes down whether it was right.
Spit bucket 0
Pantry 0
Inside the skull
The wiring from thought to reaction is not fixed. Every time a call gets checked, dopamine rewires the synapses that voted for it. The weave below is those synapses, re-knit live.
The weave
38 thinking neurons × 4 reactions. Each stitch is one learnable synapse; brighter pulls harder. Watch it change as the llama gets things right and wrong.
The senses
Fourteen sensory neurons, each one fed by a real number from the pool.
The quipu
Inca record keepers tied knots in cords. The llama keeps its score the same way: one cord per checked call, one knot per 5% the price moved, tipped green if it called it right and red if not.
How it thinks
Everything runs in your browser. Your tab has its own llama, and it remembers what it learned the next time you visit.
Graze
Every minute it pulls fresh and trending Solana pools from GeckoTerminal and newly listed profiles from DexScreener.
Sniff
It picks one it hasn't tasted. The pool's numbers become currents into 14 sensory neurons, which fire harder the stronger the signal.
Chew
48 hidden neurons (Izhikevich model, 1 ms steps) argue it out: a tasty camp, a rotten camp, a doubt camp that fires when both shout. Four motor neurons race to the line.
Learn
Four minutes later it checks the price. Right call: the synapses that voted get stronger. Wrong call: they get weaker. That is the whole trick.
The llama researches the whole chain so you don't have to. It still spits on most of it.
CA posted at launch