LAS ingestion
Upload, organize, keep per-user isolation.
Drop in LAS well logs, group them by project, and work from a single inventory. Every user’s data stays isolated while the interface stays a plain browser tab.
reservoirAI is a browser‑based workspace for repairing, analyzing, and ranking wells from digital log data. Upload LAS files, let the models repair bad‑hole sections, compare original against corrected curves, and walk out with a prioritized set of targets instead of a folder of guesses.
A tool built around one workflow — because the gap between a raw log and a defensible target list is where the money actually gets spent.
Upload, organize, keep per-user isolation.
Drop in LAS well logs, group them by project, and work from a single inventory. Every user’s data stays isolated while the interface stays a plain browser tab.
Machine learning where the caliper lied.
Detects bad‑hole intervals automatically, then reconstructs the affected curves with trained per‑curve models — GR, resistivity, density, neutron, and the rest — so washouts stop poisoning your averages.
See exactly what the model changed.
Side‑by‑side log tracks make the repair auditable: original curve, reconstructed curve, and the repair mask, all on the same depth scale.
Formation tops and pay zones, highlighted.
Mark formation tops, flag hydrocarbon zones, and generate cross‑plots for the petrophysical interpretation you’d otherwise assemble by hand.
Priority lists with ROI‑style metrics.
Turn per‑well analysis into a ranked program: which wells first, what the uplift looks like, and why — exported as reports you can hand to someone who wasn’t in the tool.
Flask, SQLite, and a machine that isn’t rented.
Originally deployed on EC2. Now self‑hosted behind Cloudflare Tunnel on our own hardware — same application, no monthly infrastructure bill, data that never leaves the lab.