Dispatches from the bench
Field notes.
Every dispatch is a real run against open data. The agent does the work, the paper keeps the score, and the difference between them is published in plain sight.
Open-source first
We pick studies whose plates we can re-run. No cherry-picking, no synthetic toy data without saying so.
Published comparisons
Every dispatch lists what the paper got, what Gradiance got, and the percentage difference. No spin.
Receipts attached
Each piece links to the source data, the model used, and a Python script that reproduces the fit.
The archive.
Browse every dispatch. Each row is a real run, with the Δ-vs-paper printed where it can't be hidden.
In the queue.
What's on the bench next. Have a published assay you'd like to see re-run? Drop us a line.
- № 05
BSA standard curve, Bradford vs BCA
Quantificationsource · Pierce/Thermo BCA insert curves · figshare BSA dilution seriesJun 2026 - № 06
Saturation binding to fit Kd and Bmax
Affinitysource · PDSP K_i database (UNC) · IUPHAR/BPS pharmacologyJul 2026 - № 07
Antiviral CC₅₀/EC₅₀ selectivity index
Cell-basedsource · PubChem BioAssay · ChEMBL antiviral compound depositsAug 2026 - № 08
Bacterial growth-curve doubling time
Kineticssource · Zenodo deposits with raw OD₆₀₀ traces (search: "E. coli growth curve")Sep 2026
Colophon
Run your own field note.
Drop a CSV, get a fit and a script that reproduces it. Three runs a month, no card.