Public datasets and environments, end to end -- ingest with a real API key, query, and check the result yourself. Nothing here is a canned demo; every number is something you can recompute independently.
NASA's CMAPSS FD001 turbofan dataset with every sensor identifier stripped -- ingest it, query it, and reproduce the blind channel-classification result against your own independent check.
Bootstrap(api_key=...) ingest/queryUCI's Occupancy Detection dataset, with a real occupancy label on every row -- check whether the local channel classifier and outcome-tagged retrieval both agree with the ground truth.
No file to download -- a deterministic Gymnasium environment run on your own machine. Bias action choice toward what previously worked, then check that against a random-policy baseline and an independent local recomputation.
pip install gymnasiumUCI's Daily and Sports Activities dataset, real wearable IMU data across 8 subjects -- reproduce the zero-config classification result yourself, including a real pooling mistake that gave a physically backwards answer before the per-subject fix.
Three independent systems with no sensor stream involved -- quantum error correction, a lattice field theory Monte Carlo simulation, and a transverse-field Ising model -- reproduced end to end, including an honest limitation and the legitimate fix that didn't change it.
A simulated drone wired into the vaas-x ingest/query/outcome loop in real time, plus a controlled ablation that found a genuine, reproducible limitation in how memory competed for the decision -- and why.
A trained encoder turning raw 6-axis IMU windows into embeddings that separate real human activities by cosine similarity -- validated on subjects held out of training, with an honest per-activity breakdown of where it's strong and where it's weak.
A frozen, untrained image embedding separating real defects from normal parts across three industrial domains -- no training, no labelled dataset for your product line. Includes the dataset-license check that ruled out two other candidates first.
The practical follow-on to the Industrial whitepaper: connecting your own machine fleet -- including a resource-constrained ESP32 sensor node -- closing the outcome feedback loop, querying across devices, and sizing a deployment.
Prefer to just try it? Get an API key and query your first episode in under ten minutes.