In upstream operations, production engineers make hundreds of decisions every week that directly influence production performance. Their objective is straightforward: identify underperforming wells, determine the cause, initiate corrective action, and ensure the loss is properly categorized.

The challenge is that modern assets generate more exceptions than any engineering team can investigate in a day.

Large outages naturally receive immediate attention. A compressor trip, facility shutdown, or major equipment failure quickly rises to the top of the priority list. The real challenge lies elsewhere, in the smaller production losses that occur across dozens or hundreds of wells every day.

Individually, these losses appear insignificant.

Collectively, they can represent substantial deferred production and a hidden source of operational inefficiency.

At Infosys Consulting, we believe the next generation of digital transformation in upstream operations will not be defined by monitoring systems alone. It will be driven by intelligent diagnostic capabilities that help engineers understand production losses faster, at greater scale, and with higher consistency.

This vision inspired Prod PILAR (Production Intelligence for Loss Attribution & Recovery), an agentic diagnostics concept designed to expand engineering coverage across entire well portfolios.

The Reality of Production Loss Attribution

Production engineers operate under three constraints that exist simultaneously: time, materiality, and coverage.

Time is finite. Experienced engineers spend a significant portion of their day investigating production anomalies, retrieving operational data, reviewing trends, and building the context required to reach a conclusion. Much of this effort is consumed by data gathering rather than analysis itself.

Materiality creates another challenge. A modest production shortfall from a single well may not justify immediate investigation. However, when similar losses occur repeatedly across multiple wells over an extended period, the cumulative business impact can become substantial.

The most significant limitation, however, is coverage.

A large upstream asset may contain hundreds of producing wells. Even with disciplined daily reviews, engineering teams can only investigate a fraction of the portfolio. As a result, many lower-priority deviations remain unexplored, not because they lack value, but because the available hours are exhausted before every well can be reviewed.

The outcome is understandable. Engineers focus on the most urgent issues while long-tail production losses receive limited attention or are assigned broad classifications during later reporting cycles.

Rethinking the Diagnostic Process

What if every production deviation received an initial assessment?

Rather than waiting for manual investigation, an intelligent diagnostic agent can continuously analyze well performance, compare actual production against expected targets, correlate operational signals, and generate a reasoned hypothesis regarding the likely source of production loss.

The objective is not to replace engineering judgment.

The objective is to ensure that every well receives attention.

Prod PILAR follows this principle by evaluating production deviations and generating a proposed loss classification supported by evidence, confidence scoring, and documented reasoning. Where sufficient evidence exists, the system provides a recommended attribution. Where evidence is incomplete, the system explicitly identifies uncertainty rather than forcing an unsupported conclusion.

This creates something many production organizations currently lack: complete coverage.

Instead of hundreds of unexplained well-days, operators gain a structured, searchable, and auditable record of production loss hypotheses across the entire asset.

From Visibility to Decision Intelligence

The energy industry has invested heavily in digitization over the past decade. Sensors, historians, dashboards, and analytics platforms have dramatically improved operational visibility.

Yet visibility alone does not solve production problems.

Engineers still need to determine why production losses occurred and where limited attention should be focused.

Agentic diagnostics represent the next step in that evolution.

By continuously evaluating production data and prioritizing wells based on confidence, impact, and operational risk, engineering teams can concentrate their expertise where human judgment creates the greatest value. Instead of spending valuable hours identifying which wells require investigation, engineers can focus on validating, refining, and acting upon the most important findings.

The result is a more scalable operating model where technology expands the reach of engineering expertise without compromising governance or accountability.

The Road Ahead

While Prod PILAR is currently a working demonstration, the concept highlights a broader opportunity for upstream operators.

As production systems become increasingly connected and data-rich, organizations have an opportunity to move beyond passive monitoring toward intelligent operational diagnosis. The ability to explain production losses consistently, rapidly, and across entire portfolios has the potential to improve production visibility, strengthen operational decision-making, and uncover value that often remains hidden within the long tail of daily operational deviations.

At Infosys Consulting, we see agentic diagnostics as an important step toward the future of intelligent operations. The goal is not simply to collect more data. The goal is to transform data into actionable insight that helps engineers make better decisions, faster.

Because in upstream operations, every deferred barrel has a story.

The question is whether anyone has the time to find it.

Please feel free to reach out to Srinivasan Santhanam – srinivasan.s23@infosys.com 

Srini Santhanam is a technology and domain consultant with over two decades of experience in the upstream oil and gas sector. He began his career working with Schlumberger’s suite of products, developing deep expertise in drilling solutions across the well lifecycle. He brings strong technical acumen in well log formats such as LAS and DLIS, having authored a custom parser for DLIS files. In the latter half of his career, he has expanded into master data management solutions, consulting for a diverse range of oil and gas companies, from mid-sized operators to global majors such as Hunt oil, Pioneer,  bp and ConocoPhillips