Dmitry Goryunov
CEO in
20+ years in oil and gas data, AI, enterprise transformation and delivery; former Accenture Oil & Gas partner and BCG X AI practice leader.
FieldState Turns Brownfield Data Into Scalable Reservoir Surveillance
A reviewable workflow for fragmented production data, legacy documents and recurring QA/QC, from production baseline to reservoir evidence.
01FFIELDSTATE
Asset review
Candidate detail
Economics
Evidence gates
Report
Your continuous production review
Recover value from production data you already have.
FieldState reviews every brownfield well on a recurring cadence and turns production signals into an evidence-backed action backlog: monitor, fix data, validate, or approve intervention.
See the asset-level backlog first, then drill into the well-level evidence.
Each candidate is ranked with screening-level economics before your team spends field or service dollars.
Your cadence
Weekly signals. Monthly decisions.
Your evidence
Concise at the top. Verifiable underneath.
FieldState Monthly Asset Review
March 2026 · 126 wells reviewed · Brownfield production review
Review cycle complete
Reviewed126 / 126 wellsFull asset coverage this cycle
Opportunity surfaced$1.8M18 wells · screening NPV
Value at risk$540k15 wells with unresolved issues
Deferred production12.8k bbl7 validation cases
All 126
Execute if validated 3
Validate hypothesis 7
Data fix 11
Ready for service handoff 3 wells · $390k
Ranked candidate list
Sorted by decision value and evidence readiness
WellBucketSignalValue
C-018Execute if validatedRestart underperf.$126k NPV
A-104Validate hypothesisDecline shift$74k NPV
B-221Data fixActive-day gap—
D-077MonitorOn trend$8k NPV
Selected candidate
Well C-018
Basic NPV$126k
IRR51%
Economic limit dateSep 2028
Evidence gradeA-
Required validation
Confirm service scope and field constraint before approval.
Your evidence
Concise at the top. Verifiable at the bottom.
Leadership gets the conclusion first. Engineers can drill into source data, formulas, segments and proxy assumptions before trusting the recommendation.
Top
Business conclusion
Recommended stance, value, urgency, owner and reason the prior read no longer applies.
Evidence
Source lineage
Production coverage, data-quality gates, selected intervals and excluded periods stay visible.
Math
Formulas and segments
DCA fits, economic inputs, segment choices, Basic NPV, IRR and confidence measures sit beside the result.
Assumptions
Proxy assumptions
Directional phase and rel-perm proxies are marked as proxies, not reservoir-model claims.
Drill-down
Calculation check
Technical pages let the operator trace the recommendation back to the underlying numbers.
Reliable Interpretation Starts With Reliable Data
Before any reservoir surveillance can be trusted, two interdependent workstreams have to move together: the data foundation that makes information usable, and the engineering interpretation built on top of it.
Data foundation and tools
Make structured and unstructured brownfield data usable without repeated manual collection, checking and re-formatting.
- Scan, read and recognize legacy documents
- Extract tables, facts, events and engineering signals
- Normalize wells, rates, events and identifiers
- Create shared analytical storage with traceable evidence
Engineering interpretation
Turn prepared and incoming data into a reviewed view of asset behavior that can be refreshed as conditions change.
- One-time interpretation of existing data
- Continuous interpretation as new data arrives
- Operational event and intervention effects
- Pressure, injection and connectivity signals
The bottleneck is not only fragmented data. It is the manual effort required to make that data usable every time the question changes.
02Mature Assets Need Continuous Engineering Capacity, Not a One-Off Interpretation
A conventional service project can deliver a one-off interpretation. The harder problem is keeping interpretation accurate while assets keep changing, manual work accumulates and engineering capacity stays limited.
Point-in-time snapshot
A one-off report explains the data available today, then starts aging as soon as the asset changes.
- Production baseline
- Initial event chronology
- Reservoir questions and data gaps
Manual change load
Even existing assets keep changing through rates, events, interventions, pressure updates and operating constraints, which creates repeated preparation and review work.
- More trends to monitor
- More assumptions to refresh
- More evidence to review
Scaled engineering capacity
Continuous workflows help the same experts cover more fields while improving accuracy and reducing repeated manual rework.
- Portfolio-level scaling
- Reusable review logic
- More current decisions
The value is not only interpretation quality. It is the ability to keep that quality current across more assets without scaling manual effort linearly.
03Five Phases Turn Manual Data Work Into Repeatable Decision Gates
Each phase turns scattered inputs and manual checks into a concrete output and a decision gate. The timing reflects the current overlapping roadmap rather than a sequential waterfall.
Phase
Inputs
Core work
Output
Decision enabled
Window
Phase 1Production Intelligence
- Rates, days-on, well metadata
- Legacy docs, scans, spreadsheets
- OCR / extraction + data structures
- Normalize, QA/QC, regime detection
- DCA, NPV / IRR, economic limit
- Canonical production baseline
- Do-nothing forecast
- First economics and quality gate
- Which history is trusted
- What the asset does today
- How long it lives as-is
M1-M6
Phase 2Operational Intelligence
- Phase 1 baseline
- Workovers, uptime, choke settings
- Operating event history
- Event chronology and causal model
- What-if scenarios
- Deferred production / loss logic
- Operational event layer
- Scenario workspace
- Opportunity-loss cashflow view
- Which levers matter
- Expected uplift or loss
- What to test first
M4-M10
Phase 3Reservoir Surveillance
- Phase 2 + pressure data
- Injection history, well tests
- Connectivity evidence
- Pressure / injection intake
- RTA workflow
- Pressure support and connectivity validation
- Pressure-aware reservoir behavior
- Connectivity map
- Field loss / recovery impact view
- Where support is evidenced
- Where uncertainty remains
- What recovery risk to address
M7-M14
Phase 4Integrated Optimization
- Phase 3 + well models
- Equipment limits, surface constraints
- SCADA / historian / workover path
- Nodal / VLP logic
- Constraint and bottleneck model
- Integration path to operations
- Constraint-aware workspace
- Daily optimization view
- System bottleneck map
- Which actions are feasible
- Which constraints block value
- What operations can execute
M9-M18
Phase 5Agentic Engineering
- Validated rules and constraints
- Review gates, workflows, evidence packs
- Prior phase outputs
- DAG workflow orchestration
- Selective AI reasoning
- Audit trail and exception routing
- Reusable engineering workflow
- Review logic and traceability
- AI-assisted daily workflow
- Which decisions repeat
- What can be reviewed safely
- What can scale across assets
M11-M18
Delivery roadmap
High-Level Delivery Roadmap
Each bar represents a full product phase. The roadmap shows how the five workstreams overlap instead of running as a sequential waterfall.
M1M2M3M4M5M6M7M8M9M10M11M12M13M14M15M16M17M18
Production Intelligencebase-case operating picture
Phase 1 · Production Intelligence
Operational Intelligenceevents and scenario layer
Phase 2 · Operational Intelligence
Reservoir Surveillancereservoir behavior and uncertainty
Phase 3 · Reservoir Surveillance
Integrated Optimizationsystem optimization and constraints
Phase 4 · Integrated Optimization
Agentic EngineeringAI-assisted workflows and review
Phase 5 · Agentic Engineering
Data Needed by Phase
The discussion should confirm what data exists, what can be extracted quickly, and which later phases depend on sources that may not be ready on day one.
1. Production Intelligence
Minimum useful starting point.
- Oil, water, gas rates
- Days-on production
- Basic well metadata
- Legacy files for extraction
2. Operational Intelligence
Turns baseline into controllable scenarios.
- Choke settings
- Uptime and downtime
- Workover / intervention history
- Event chronology
3. Reservoir Surveillance
Moves from production behavior to reservoir behavior.
- Pressure data
- Injection history
- Well tests
- Connectivity evidence
4. Integrated Optimization
Connects engineering constraints to operations.
- Well models
- Equipment limits
- Surface constraints
- SCADA / historian path
5. Agentic Engineering
Requires validated workflows before AI can assist safely.
- Engineering rules
- Approval constraints
- Workflow history
- Review evidence packs
AI Removes Manual Work Around the Engineering Process
Document and table extraction
Legacy reports, scans and spreadsheets are converted into structured, source-linked facts.
Normalization and QA/QC
Wells, dates, units, events and gaps are checked before engineers spend time on interpretation.
→
Evidence packs for review
Assumptions, confidence and source references are prepared for SME review.
Workflow automation
Repeatable phase tasks and review gates become traceable workflows instead of ad hoc manual work.
Continuous refresh
New incoming data updates the surveillance state without rebuilding the whole interpretation from scratch.
AI is used around the engineering workflow: preparation, checking, evidence and refresh. Engineers still approve interpretation and decisions.
08The First Discussion Should Confirm Where Manual Data Work Blocks Surveillance
We do not need to price the full program in the first meeting. We should confirm the asset scope, available data, manual bottlenecks and the first useful decision gate.
1
Asset scope
Fields, wells, systems and the surveillance priority you want to address first.
2
Available data
Production exports, legacy documents, event history, pressure, injection and well-test evidence.
3
Manual bottlenecks
Where engineers spend time collecting, checking, reconciling or explaining data today.
4
First decision gate
The first phase output that would be useful enough to plan in detail with your team.
Why this team
A team already operating as one unit from ADNOC ENERGYai.
Shared delivery proof
We worked together on the ADNOC ENERGYai program in Abu Dhabi: same enterprise energy environment, same trust bar, same need for repeatable engineering evidence.
The chemistry is already tested: we know how to set goals, divide ownership, challenge each other, and turn complementary skills into shipped work.
Artem Gendler
COO / Product in
15+ years building enterprise products and AI workflows, including 5 years in oil and gas; turns ambiguous operator problems into auditable product systems.
Alex Petrov
CPO in
17+ years across upstream engineering and EnergyAI: drilling, production optimization, reservoir behavior, field operations, late-life assets and AIQ/ADNOC-G42 systems for operational reasoning.