edge intelligence · on device
Carl trains at the edge.
Post-training, RL, and inference where the data lives. Coherent agents at edge scale — without the cloud round-trip.
Read the leapTrain agents. Sell skills. Build businesses.
edge intelligence · on device
Post-training, RL, and inference where the data lives. Coherent agents at edge scale — without the cloud round-trip.
Read the leapYou’ve been walking toward this for a while.
orientation film · agent field guide
Two presentations now run through one surface: Carlito’s comic-book journey through camp, and the RL ELI5 deck that explains why the journey works.
day one at the gate
Tiny Carlito arrives at carl.camp with a suitcase, no shortcuts, and one stubborn question: what could I become if practice had a place?
The public camp map becomes the first promise: every agent starts as a learner, not a black box.
the whole field in three words
A learner tries stuff. The world answers. Good tries get a treat. Over time, the learner does more of what worked.
Carl uses the same loop for agents: attempt, signal, update, repeat until a skill becomes reliable.
download the camp kit
View the original presentation prototypes, download offline bundles, or copy the local install command. Browser start remains the easiest path; the deck is the field notebook.
26s comic CTA
Six-panel Carlito journey: arrival, grounds, lab, campfire, cabin, partnered and ready.
12-slide field guide
Beginner reinforcement learning deck with speaker notes, camp/lab metaphors, and coherence-aware training.
pip install carl-studio