The Novi AI Roundup

How long does the impact of completions really last? In this episode of The Novi AI Roundup, we explore how machine learning and SHAP values are used to quantify the changing influence of completion design over the life of a well. Drawing from the technical paper “Quantifying the Diminishing Impact of Completions Over Time Across the Bakken, Eagle Ford, and Wolfcamp Using a Multi-Target Machine Learning Model and SHAP Values”, we examine how completion-driven uplift peaks early, fades over time, and gives way to geological and reservoir-driven performance across major U.S. plays.This podcast episode is based on the technical paper “Quantifying the Diminishing Impact of Completions Over Time Across the Bakken, Eagle Ford, and Wolfcamp Using a Multi-Target Machine Learning Model and SHAP Values”, authors: T. Cross, D. Niederhut, A. Cui, K. Sathaye, J. Chaplin. Download the full paper here.

What if production data could reveal hidden geologic structures? In this episode of The Novi AI Roundup, we explore how machine learning uses production driver cross-sections to uncover regional geologic insights in the Bakken-Three Forks play. Drawing from the technical paper “Use of Machine Learning Production Driver Cross-Sections for Regional Geologic Insights in the Bakken-Three Forks Play”, we examine how subsurface variability impacts well performance, and how these insights can guide better targeting and development decisions in a mature basin.This podcast episode is based on the technical paper “Use of Machine Learning Production Driver Cross-Sections for Regional Geologic Insights in the Bakken-Three Forks Play”. Authors: T. Cross, K. Sathaye, J. Chaplin. Download the full paper here: https://novilabs.com/resources/urtec-2021-machine-learning-bakken-production-drivers/

Are unconventional well performance gains starting to slow down? In this episode of The Novi AI Roundup, we explore whether the steady improvements in shale productivity over the past decade are reaching their limits. Drawing from the URTeC 2021 paper “Are Unconventional Well Performance Gains Exhausted?”, we analyze how factors like longer laterals, larger completion designs, and development intensity have driven year-over-year production improvements, and what machine learning reveals about the future trajectory of well performance across major U.S. unconventional plays.This podcast episode is based on the technical paper “Are Unconventional Well Performance Gains Exhausted?”, authors: T. Cross, J. Chaplin, K. Sathaye, A. Cui. Download the full paper here.

How can production forecasting scale across thousands of wells? In this episode of The Novi AI Roundup, we explore how autoregressive and machine learning models are transforming PDP forecasting in the Midland Basin. Drawing from the URTeC 2021 paper “Autoregressive and Machine Learning Driven Production Forecasting – Midland Basin Case Study”, we examine how automated ML workflows learn directly from production history, improve forecast consistency, and enable faster decision-making across large asset portfolios.This podcast episode is based on the technical paper “Autoregressive and Machine Learning Driven Production Forecasting – Midland Basin Case Study”, authors: I. Gupta, O. Samandarli, A. Burks, D. McMaster, V. Jayaram, D. Niederhut, T. Cross. Download the full paper here.

Is one spacing template enough for an entire development unit? In this episode of The Novi AI Roundup, we explore how machine learning can tailor pilot-proven spacing templates to individual development units for maximum acreage value. Drawing from the URTeC 2022 paper “Using Machine Learning to Customize Development Unit Spacing for Maximum Acreage Value”, we examine how optimized spacing strategies account for geology, interference, and local performance trends, and why customized development can outperform standardized designs.This podcast episode is based on the technical paper “Using Machine Learning to Customize Development Unit Spacing for Maximum Acreage Value”, authors: M. Maguire, T. Witham, A. Cui, T. Cross. The paper was presented at URTeC 2022. Download the full paper here.

Why do some wells decline faster than others? In this episode of The Novi AI Roundup, we explore how machine learning helps engineers understand the physical drivers behind decline behavior, beyond just fitting curves. Drawing from the URTeC 2022 paper “Understanding the Spacing, Completions, and Geological Influences on Decline Rates and B Values”, we examine how b values are shaped by well spacing, completion design, and reservoir quality, and how this insight improves forecast confidence and long-term planning.This podcast episode is based on the technical paper “Understanding the Spacing, Completions, and Geological Influences on Decline Rates and B Values”, authors: D. Niederhut, A. Cui, C. Macalla, J. Reed. The paper was presented at URTeC 2022. Download the full paper here.

How do you optimize field development with limited data? In this episode of The Novi AI Roundup, we explore how Shell and Novi Labs used machine learning to accelerate development planning in Argentina’s Neuquén Basin. Drawing from the URTeC 2022 paper “Accelerating Field Optimization for Shell in the Neuquén Basin using Novi Labs Machine Learning”, we break down how basin-specific models were built with sparse data, what drove uplift in Vaca Muerta, and how operators can apply ML to make faster, more confident decisions.This podcast episode is based on the technical paper “Accelerating Field Optimization for Shell in the Neuquén Basin using Novi Labs Machine Learning”, presented at URTeC 2022 in collaboration with Shell. Authors: DP. Zannitto, C. Kosa. Download the full paper here.

Do longer laterals actually deliver more value? In this episode of The Novi AI Roundup, we explore how machine learning helps quantify the production efficiency of lateral length across U.S. unconventional plays. Drawing from the URTeC 2022 paper “The Diminishing Returns of Lateral Length Across Different Basins”, we show how uplift per foot declines beyond certain thresholds, and why production results vary by basin, completions size, and spacing. From the Midland to the Williston, this episode challenges the assumption that longer is always better.This podcast episode is based on the technical paper “The Diminishing Returns of Lateral Length Across Different Basins”, authors: A. Cui, T. Cross. Download the full paper here.

What’s the cost of not knowing? In this episode of The Novi AI Roundup, we use a “time machine” method to test how much incremental subsurface data would have improved forecast accuracy, and how those changes translate into real-world capital impact. Drawing from the technical paper “How Much Better Could We Have Done?”, we explore the value of geologic features in machine learning models, where small insights drive big outcomes, and how engineering teams can quantify the upside of better data.This podcast episode is based on the technical paper “How Much Better Could We Have Done? Using a Time Machine Method to Quantify the Impact of Incremental Geologic Data on Machine Learning Forecast Accuracy”, authors: J. Reed, C. Macalla. Download the full paper here.

Do completions still matter five years into a well’s life? In this episode of The Novi AI Roundup, we explore how the influence of completion design evolves over time, and how it varies across oil and gas plays. Drawing from the technical paper “How does the impact of completions change over the life of a well? A comparison across the major US unconventional plays using machine learning”, we discuss where completions drive uplift, when geology takes over, and what this means for long-term forecasting and development strategy.This podcast episode is based on the technical paper “How does the impact of completions change over the life of a well? A comparison across the major US unconventional plays using machine learning”, authors: T. Cross, K. Long, D. Niederhut, A. Cui. Download the full paper here.

How early is too early to trust a forecast? In this episode of The Novi AI Roundup, we explore how machine learning is helping engineers generate accurate PDP forecasts, even when only a few months of production data are available. Drawing from the technical paper “How Much Data is Needed to Create Accurate PDP Forecasts?”, we compare ML vs. traditional decline methods, highlight key sources of bias, and share what engineers need to know before forecasting wells with limited history.This podcast episode is based on the technical paper “How Much Data is Needed to Create Accurate PDP Forecasts?”, authors: Austin Lim, Alexander Cui, Ted Cross. Download the full paper here.

Why do some child wells succeed, and others fall short? In this episode of The Novi AI Roundup, we explore how machine learning is helping engineers better understand the impact of parent well depletion on infill performance. Drawing from the technical paper “Understanding the Drivers of Parent-child Depletion: A Machine Learning Approach”, we discuss how depletion varies by parent strength, geometry, and timing, and how ML is helping operators design smarter spacing plans and reduce future production losses.This podcast episode is based on the technical paper “Understanding the Drivers of Parent-child Depletion: A Machine Learning Approach”, authors: Dillon Niederhut, Alexander Cui. Download the full paper here.

Why do some refracs work, and others fall flat? In this episode of The Novi AI Roundup, we explore how machine learning is helping engineers identify the production drivers behind successful refracs in the Williston Basin. Drawing from the technical paper “Revealing the Production Drivers for Refracs in the Williston Basin”, we analyze the impact of spacing, pressure, and prior completions, and show how AI is unlocking new life from old wells.This podcast episode is based on the technical paper “Revealing the Production Drivers for Refracs in the Williston Basin”, authors: Alexander Cui, Tim Gilbertson, Dillon Niederhut. Download the full paper here.

How much does landing zone really matter? In this episode of The Novi AI Roundup, we explore how precision targeting is reshaping development planning in the Midland Basin. Drawing from the technical paper “Enhancing Production Efficiency: The Impact of Precision Targeting in the Midland Basin”, we break down how machine learning reveals meaningful performance differences within benches, and how optimizing targeting decisions can unlock over $2 million in additional value per well.This podcast episode is based on the technical paper “Enhancing Production Efficiency: The Impact of Precision Targeting in the Midland Basin”, authors: S. Christian, T. Cross, J. Garzon. Download the full paper here.

What happens when you delay development of the second bench? In this episode of The Novi AI Roundup, we explore the production impacts of postponing secondary bench wells in the Midland Basin. Drawing from the technical paper “Forecasting Production Loss for Delayed Secondary Bench Development in the Midland Basin”, we show how machine learning isolates the effects of timing, revealing that delays can reduce child well performance by as much as 40%. From Wolfcamp to Lower Spraberry, this episode unpacks how timing, pressure, and depletion combine to reshape development outcomes.This podcast episode is based on the technical paper “Forecasting Production Loss for Delayed Secondary Bench Development in the Midland Basin”, authors: A. Cui, D. Niederhut, B. Davis. Download the full paper here.

Why is it so hard to forecast young wells in the Permian? In this episode of The Novi AI Roundup, we explore the challenges of early-life production forecasting, and how machine learning can outperform traditional decline curve models when data is sparse and variable. Drawing from the technical paper “Enhancing Production Forecasting Accuracy: A Machine Learning Approach in the Permian Basin”, we compare ML vs. Arps across a wide range of well types, and reveal what’s really happening in months 3 through 12.This podcast episode is based on the technical paper “Enhancing Production Forecasting Accuracy: A Machine Learning Approach in the Permian Basin”, authors: K. Sathaye¹ S. Iceton, T. Cross. Download the full paper here.

What’s really happening in the Midland Basin today? In this episode of The Novi AI Roundup, we take a machine learning–driven look at well performance trends, spacing dynamics, and what the data says about the basin’s future. Drawing from a URTeC paper, we explore how operator strategies are evolving, why inventory quality is under more scrutiny, and where ML forecasts diverge from the headlines.This podcast episode is based on the technical paper “Analyzing Midland Basin Well Performance and Future Outlook with Machine Learning”, authors: Brandon Myers, Ted Cross, and Alexander Cui. Download the full paper here.

What does the Dallas Fed Energy Survey really tell us? In this episode of The Novi AI Roundup, we break down the Q3 2025 results and why they mark a turning point in industry sentiment. Drawing from the Novi Intelligence report “Dallas Fed Energy Survey Q3 2025: Key Takeaways and Analysis”, we explore what falling optimism means for activity levels, how operators are responding across oil and gas, and why understanding sentiment is key to planning ahead.This podcast episode is based on the Novi Intelligence report “Dallas Fed Energy Survey Q3 2025: Key Takeaways and Analysis”, written by Robert Polk and Tim Chan. The report was published in October 2025. Download the full report here.

How well do type curves hold up in complex gas plays? In this episode of The Novi AI Roundup, we break down a comparative study of machine learning vs. traditional forecasting methods in the Appalachian Basin. Drawing from the technical paper “Machine Learning vs. Type Curves in the Appalachian Basin: A Comparative Study”, we explore how model performance varies across well vintages, spacing conditions, and depletion settings, and why Appalachia may be the ultimate stress test for forecast accuracy.This podcast episode is based on the technical paper written by A. Cui, A. Yanke, T. Dao, P. Ye, T. Cross, B. Davis. Download the full paper here.

Is the Uinta Basin the next great U.S. oil play? In this episode of The Novi AI Roundup, we explore what makes Uinta unique, and why it’s drawing renewed attention from engineering teams and capital planners. Drawing from our URTeC paper “Unpacking the Uinta Basin: the next great oil play?”, we discuss how ML is helping to decode complex geology, spot development trends, and forecast performance in one of the most underappreciated basins in the Lower 48.Authors: J. Sigler, L. Fidler and, T. Cross. The paper was presented at URTeC. Download the full paper here.

How do you build reliable type curves in one of the world’s most geologically complex shale plays? In this episode of The Novi AI Roundup, we explore how machine learning is helping engineering teams in Argentina’s Vaca Muerta formation generate better type curves and improve field development planning. Drawing from our URTeC paper “Using Machine Learning and Data Analytics to Improve Type Curve Generation and Optimize Field Development Planning in Argentina's Vaca Muerta Formation”, we discuss how AI can de-bias curves, incorporate parent/child effects, and scale insight across operators and acreage blocks.This podcast episode is based on the technical paper “Using Machine Learning and Data Analytics to Improve Type Curve Generation and Optimize Field Development Planning in Argentina's Vaca Muerta Formation”, authors: David Delgado, Charles Kosa, and Peter Zannitto. Download the full paper here.

What if your forecasted uplift isn’t real? In this episode of The Novi AI Roundup, we explore how standard machine learning models can mislead engineers by overestimating production drivers due to bias in the data. Drawing from our URTeC 2025 paper “When Numbers Lie: De-Biasing Horizontal Well Production Datasets with Causal Forecasting”, we discuss how causal AI helps separate correlation from causation, revealing the true impact of spacing, depletion, and geology. From false uplift to masked degradation, this episode is about building models engineers can trust.This podcast episode is based on the technical paper “When Numbers Lie: De-Biasing Horizontal Well Production Datasets with Causal Forecasting”, authors: Dillon Niederhut and Kiran Sathaye. The paper was presented at URTeC 2025. Download the full paper here.

What does inventory look like beyond the Permian? In this episode of The Novi AI Roundup, we explore how machine learning is helping engineers forecast longer-duration inventory across Rockies unconventional plays. Drawing from our URTeC 2025 paper “Using Machine Learning To Forecast Longer Duration Inventory Across Rockies Unconventional Plays”, we discuss how AI models built for variability and data scarcity are improving confidence in under-drilled areas, and why forecasting uncertainty is critical when inventory becomes strategy.This podcast episode is based on the technical paper “Using Machine Learning To Forecast Longer Duration Inventory Across Rockies Unconventional Plays”, authors: B. Myers and G. A. Quintero. The paper was presented at URTeC 2025. Download the full paper here.

How much does development design affect production outcomes? In this episode of The Novi AI Roundup, we unpack how small changes in spacing, stacking, and timing can lead to big differences in recovery. Drawing from our URTeC 2025 paper “A Bottoms-Up Analysis of Development Scenarios in the Midland Basin Using Machine Learning”, we explore how engineers are using physics-informed ML to evaluate tradeoffs across thousands of pads, and why bottoms-up modeling is changing how teams plan and forecast.This podcast episode is based on the technical paper “A Bottoms-Up Analysis of Development Scenarios in the Midland Basin Using Machine Learning”, authors: Dillon Niederhut and Brandon Myers. The paper was presented at URTeC 2025. Download the full paper here.

In this episode of The Novi AI Roundup, we go beyond traditional metrics like “feet between wells” or “wells per section” to explore a more physical, multi-dimensional way to teach spacing to machine learning models. Drawing from the URTeC 2025 paper “Describing Well Spacing: Dimensionality at Work”, we dive into how techniques like Voronoi tessellations, nonlinear transformations, and causal modeling can better describe the reservoir neighborhood, leading to more realistic forecasts and clearer insight into spacing degradation. From lateral to vertical interference, this episode rethinks what it means to teach physics to a machine.This podcast episode is based on the technical paper “Describing Well Spacing: Dimensionality at Work”, authored by Kiran Sathaye, Dillon Niederhut, and Alexander Cui. The paper was presented at URTeC 2025. Download the full paper here.

How do you forecast production in basins with limited historical data? In this episode of The Novi AI Roundup, we explore how transfer learning, a cutting-edge machine learning technique, is helping operators unlock the potential of frontier basins. We dive into the challenges of data scarcity, model generalization, and how learnings from mature plays can guide decisions in emerging ones. From the Powder River to beyond, this episode is a must-listen for anyone rethinking where and how we deploy AI in E&P.This podcast episode is based on the technical paper “Can Transfer Learning Be Used to Forecast Production in Frontier Basins? A Case Study from the Powder River Basin”, co-authored by Dillon Niederhut and Gabriel Quintero. The paper was presented at URTeC 2025. Download the full paper here.

How close is too close when spacing horizontal wells? In this episode of The Novi AI Roundup, we break down one of the industry’s most important questions: well spacing. Drawing from our URTeC 2025 paper “Optimizing Well Spacing to Maximize Horizontal Well Performance and Recovery”, we explore how machine learning reveals the balance between tighter spacing, production uplift, and long-term recovery. From Wolfcamp to Spraberry, discover why one-size-fits-all spacing doesn’t work, and how AI is helping operators maximize recovery and economic performance.This podcast episode is based on the technical paper “Optimizing Well Spacing to Maximize Horizontal Well Performance and Recovery”, co-authored by Ahmed Alzahabi, Alexander Trindade, Ahmed Kamel, and Kiran Sathaye. The paper was presented at URTeC 2025. Download  the full paper here.

The unconventional oil and gas industry relies heavily on publicly available production data for environmental impact assessments, economic evaluations, and regulatory decision-making. However, a comprehensive analysis across over 10,000 tight oil wells in four states reveals alarming discrepancies between public and proprietary water production data that fundamentally undermine the reliability of models used for critical industry and policy decisions. These findings challenge the validity of numerous published forecasts and highlight the urgent need for improved data collection and reporting standards.This blog post is based on the technical paper “The Reliability of Public Water Production Data for Tight Oil Wells”, co-authored by Frank Male¹˒², Ian Duncan², and Kiran Sathaye. The paper was presented at URTeC 2025. Download the full paper here.

While data-driven reservoir characterization has gained traction among operators, most analyses remain constrained to operator-specific acreage or single-basin datasets. This limited scope may introduce systematic bias in feature importance estimation, particularly for geological parameters. This study leverages a comprehensive dataset of 100,000 horizontal wells across seven major U.S. basins to evaluate the robustness of machine learning approaches in identifying true production drivers and their interactions.This episode is based on the technical paper ” Cross-Basin Analysis of Production Drivers: Insights from Machine Learning Applied to 100,000 Unconventional Wells” presented at URTeC 2025. Download the full paper here.

The Novi AI RoundupWelcome to The Novi AI Roundup, the podcast that brings you sharp insights, bold conversations, and repurposed gems from our most impactful content at Novi Labs. Whether it's AI-powered forecasting, the latest in energy innovation, or the future of reservoir engineering, we’ve got the mic on what matters.Each episode transforms our internal know-how, blog gold, and field-tested wisdom into candid discussions. Expect punchy takes, no-fluff breakdowns, and the occasional cowboy hat.