ML forecastingResery
Curated data & ML for recovery decisions
Resery gives upstream teams production-ready models and curated training datasets for recovery decisions. Get P10/P50/P90 EUR forecasts for existing and new wells, screen refrac candidates, rank IOR options, and estimate EOR potential — built for reservoir workflows, not generic tabular ML. API-first at launch, and consumable inside WellStrata premium plans.
Capabilities
Everything Resery does
ML Production Forecasting
The flagship service: P10/P50/P90 EUR and monthly-rate forecasts for existing and brand-new wells — from well headers, completion parameters, and formation context. Built for reservoir workflows, not generic tabular regression.
Refrac Screening
Screen refrac candidates in seconds. Models rank existing wells by expected uplift from a re-stimulation, so completion teams focus diligence on the wells most likely to respond.
IOR Selection & Ranking
Rank improved-recovery options for a well or a group of wells. The models weigh formation, completion, and production signals to surface which interventions are most promising, and in what order.
EOR Estimation
Estimate enhanced-recovery potential and compare scenarios side by side — explore how different recovery strategies change the expected outcome before committing engineering time or capital.
Curated Training Datasets
Clean, model-ready datasets curated from US onshore regulatory and production data — production, completions/FracFocus, and formations across TX, OK, ND, NM, CO, and WY. The same data that powers the models.
Reservoir-Aware Models
Models are built around upstream concepts — month-on-production alignment, lateral length, proppant loading, formation aliases — not off-the-shelf feature engineering. They speak the language of the subsurface.
Scenario Comparison
Compare recovery scenarios programmatically. Change an input, re-request, and see how the expected EUR and rate profile shift — the range of outcomes, not a single point estimate.
API-First Access
Every capability is a documented endpoint: authenticate with an X-API-Key header, POST a well, get JSON back. Consume the same models inside WellStrata premium plans or standalone.
How it works
From input to answer
Curated data from US onshore sources
Training data is curated from regulatory and production records — production histories, completions/FracFocus, and formation coverage across six states.
Models trained for the subsurface
Wells are aligned on month-on-production and enriched with completion and formation context to train models for forecasting, refrac screening, IOR ranking, and EOR estimation.
You call the API
Authenticate with an X-API-Key header, POST a well to an endpoint like /v1/forecast, and get structured JSON back — P10/P50/P90 EUR, ranked options, or scenario estimates.
Consume it in your workflow
Drop the JSON into notebooks, pipelines, and economics models — or use the same models inside WellStrata premium plans.
Ready to build with Resery?
Request API access or talk to our team about how Resery fits into your upstream workflows.
