30-Session Programme · 25 Minutes Per Day · Free · For leaders in Nigeria’s oil and gas sector
WEEK 1 OF 6 · THE INFLECTION: FROM DATA TO BARRELS · SESSIONS 1–5 · WORK AT YOUR OWN PACE
Orientation · The ThesisSESSION 1
Leadership Using AI and the Six Frameworks
Why This Matters
The AI pitch is easy to find. The harder question is which loss a better decision would recover. Over thirty sessions you will build a signed 90-Day AI Survival Game Plan for your own company, one section per session, around three things you choose today: your company, its type, and the one loss that costs it most. Then you find where your company's Operating-Intelligence Loop breaks. The book is blunt about why this comes first: if you cannot say which stage of your own loop is broken, you are not yet ready to spend one naira on a model.
25-Minute Module
0:00 – 5:00
A declined session, and the finish line
- Not long ago I offered to speak, at one of the industry's flagship gatherings, on what AI is about to do to producing oil and gas in Nigeria. The answer came back politely: a session on AI was not needed. Not hostility to technology, but something quieter and more dangerous: the settled belief that this change can be watched from the platform rather than boarded.
- The six weeks: Week 1 The Inflection: From Data to Barrels. Week 2 The Foundations You Cannot Skip. Week 3 Intelligence in the Field. Week 4 Gas, the Downstream and Operational Trust. Week 5 Controlled Autonomy, Capital and People. Week 6 The Leadership Mandate.
- The finish line is a signed 90-Day AI Survival Game Plan for your own company: twenty-nine sections written one per session, assembled and signed in Session 30.
- House rules: write about your own company; bring a real loss, decision or proposal; no value claim without a baseline; no decision without a named owner; report a blocker the same day. The pass mark is 70.
- Said once, as the book says: this is leadership training, not investment, legal or engineering advice.
The platform is not a safe place to stand for long.
5:00 – 10:00
The argument in one sentence, and why now
- THE DOCTRINE
Artificial intelligence will not save Nigerian oil and gas; leadership using artificial intelligence will.
- The same frontier systems reach an operator in Port Harcourt and a major in Houston in the same week; the cloud is rentable by the hour; the talent is reachable. If technology were the binding constraint, the problem would already be solved.
- The constraint is executive discipline: data foundation first, bounded pilots not slideware, human judgement where stakes are high, scale only what pays.
What the book cites Figure Attributed to Crude production ambition 2 million barrels per day by 2027; 3 million by 2030 NNPC Limited Fresh upstream investment ambition $30 billion to $60 billion by the end of the decade NNPC Limited 2025 production target At least 2.1 million barrels per day NUPRC Recent output Closer to 1.5 to 1.7 million barrels per day The book's introduction Gap to the 2030 ambition (recomputed) 3.0 less 1.7 = 1.3 million; 3.0 less 1.5 = 1.5 million barrels per day Arithmetic on the figures above - The second reason is measurement. In April 2026 the regulator directed upstream operators toward standardised templates and measurement-based methane and greenhouse-gas reporting, within a national framework of net zero by 2060, an end to routine flaring by 2030 and a methane-reduction target for 2035. The discipline that makes AI work is the one compliance and capital now demand: one effort, three returns.
The gap will be closed, if at all, by execution.
10:00 – 17:00
Six frameworks, one page
Framework What it says The question it forces 1. The Operating-Intelligence Loop Sense, decide, act, learn. Value appears only when the loop closes. At which stage does our loop break? 2. The Data-Foundation Precondition Garbage in, confident garbage out. Is the data behind our key decision trustworthy enough? 3. The Autonomy Ladder Tools inform, copilots draft, agents execute. Autonomy is earned rung by rung. Which rung has this system earned? 4. The Value-versus-Risk Map Value created against damage done when wrong. A wrong playlist is not a wrong wellhead call. What does it cost us when it fails, and is the value worth that exposure? 5. The Sequencing Discipline Foundation, bounded pilot, human in the loop, then scale what paid. Are we doing things in the right order? 6. The Operations-Excellence Lens Judge every change by what the business already measures. Which barrel, downtime or cost number moves? - THE DOCTRINE
Autonomy granted faster than it is governed is not innovation. It is industrialised risk.
- The failing sequence is the inverse: buy a platform, hire data scientists, build a proof of concept, and find eighteen months later that nothing in the field has changed. The lens gives one instruction: connect the capability to the barrel, or do not build it.
Perfection is not required to begin; a trustworthy enough foundation around the decisions that matter is.
17:00 – 22:00
Classify your company, choose the loss, find the break
- Classify your company by the book's own company types: upstream operator; marginal field operator; EPC and construction; drilling and well services; marine, logistics and vessel operators; fabrication yards and workshops; downstream, midstream and gas. If you span two, choose the larger. Week 6 returns to these types.
- Choose the one loss that costs your company most, in the lens's terms: barrels, downtime, decision delay, compliance exposure or cost. A number is welcome; an invented one is not.
- Walk that loss round the loop: sense through sensors, reports and records; decide by turning signals into a ranked, owned decision; act through a named owner with a deadline; learn by measuring the outcome.
- IN THE FIELD · THE WORKED CASE
- Field. A fictional marginal field operator installs a screen showing every well. Nobody ranks the wells that need attention, so the loop breaks at decide.
IN THE BOARDROOM · THE PARALLEL- Boardroom. The same fictional company's executive committee approves workovers but never checks whether the barrels came back: a break at learn. Field and boardroom can break at different stages of one loop.
- THE DOCTRINE
The loop is the unit of value. Value appears only when it closes: sense, decide, act, learn.
Name the stage that breaks, not the one you would like to buy.
22:00 – 25:00
Write it down, and Section 1
- Before you close the laptop, write three lines: company and type; the one loss; the breaking stage.
- Section 1 of your plan is Our company, its type, the one loss that costs us most, and where our Operating-Intelligence Loop breaks. Choose a loss you will still care about in Session 30.
- The book's charge: learn the six until they are reflexes, then refuse to be impressed by anything that ignores them.
If nobody acts when the signal arrives, write that down. It is your first finding.
Key Terms
Operating-Intelligence Loop
Sense, decide, act, learn. Value appears only when the loop closes.
Confident garbage out
A model fed poor data returns a fluent, well-formatted answer that is wrong and sounds certain.
Autonomy Ladder
Tools that inform, copilots that draft and suggest, agents that execute. Each rung is earned and matched to consequence.
Binding constraint
The limit that actually holds performance back. In the book's argument, executive discipline rather than technology.
Self-Check — Before You Submit
1. A fictional upstream operator's new screens show every compressor in real time. A year on, deferment has not fallen and no one is assigned to act on them. Where has its loop most likely broken?
- A. At sense: it needs more sensors
- B. At decide: signals are not converted into a ranked, owned decision
- C. At learn: the model needs retraining
- D. Nowhere yet: dashboards take time to show value
B. At decide: signals are not converted into a ranked, owned decision — Most programmes build the sensing stage and stop. More sensors is the tempting answer, but sensing is the one stage already working here.
2. According to the book, why is technology not the binding constraint on AI in Nigerian oil and gas?
- A. Nigerian operators cannot yet access frontier models
- B. Regulators restrict AI upstream
- C. The cloud is too expensive at scale
- D. Models are commoditised, the cloud is rentable and the talent is reachable
D. Models are commoditised, the cloud is rentable and the talent is reachable — If technology were the constraint, the problem would already be solved. Access to models is the tempting wrong answer; the book says the same systems reach Port Harcourt and Houston in the same week.
3. A fictional drilling and well services company plans to buy a platform, hire data scientists, then find a problem. What does the Sequencing Discipline say?
- A. Proceed; early investment signals commitment
- B. Proceed if the vendor is a major
- C. This is the failing sequence: foundation and a bounded pilot first
- D. Hire the data scientists but delay the platform
C. This is the failing sequence: foundation and a bounded pilot first — The book names this as the sequence that fails. Delaying only the platform still starts with technology rather than a measured problem.
Today’s Assignment
Name your company and its type, the one loss that costs your company most, and the stage of the Operating-Intelligence Loop where its loop breaks.
- Your company. What it does and at what scale; it need not be named. Its type from the book's seven, and why.
- The one loss. What it is, where it occurs, how it is measured today and roughly what it costs. Write not known rather than invent a figure.
- The loop walk. What happens today at sense, decide, act and learn for that loss. Name the breaking stage and the evidence.
- The honest part. What your company currently gets wrong about this loss. No AI in use yet is normal.
- Four hundred to nine hundred words.
YOUR 90-DAY PLAN
Section 1 — Our company, its type, the one loss that costs us most, and where our Operating-Intelligence Loop breaks.
How this is marked — the rubric
| Classifies the company by a book company type with a reason | 15 |
| Names one specific loss and how it is measured, marking unknowns rather than inventing figures | 30 |
| Walks all four loop stages and names the breaking stage with evidence | 35 |
| The honest paragraph names a failing in the student's own company, not the industry | 20 |
| Total · pass mark 70 | 100 |
Submitting session 1
Write your answer in Word or Google Docs. The first line must read
Write your answer in Word or Google Docs. The first line must read
Session: 1, or the submission is refused. Upload the Word file (.docx), or switch to the Google Doc tab and paste your document’s share link.Marked against the rubric above and returned as a written letter, saved to your progress page. Do not submit until you have passed the self-check above.Capabilities Covered
The book's one-sentence argumentThe six toolkit frameworksLoop diagnosisCompany classificationPlan authorship
Part One · The InflectionSESSION 2
Where the Next Barrel Comes From
Why This Matters
Every board in this sector has seen the national targets. The question Chapter 1 puts at the centre of every board discussion is harder: where, precisely, will the next barrel come from? Some of the answer is new drilling, better security and new evacuation routes. A meaningful part is cheaper and less photogenic: better performance from what is already operated. Today you value your Session 1 loss, test one AI idea against field economics, and find where your company's data stops converting into decisions.
25-Minute Module
0:00 – 6:00
Where the next barrel comes from
- Session 1 gave you the crude targets. Chapter 1 adds gas: NNPC Limited's ambition to grow gas production to 10 billion cubic feet per day by 2027 and 12 billion cubic feet per day by 2030.
- The book's sources of the next barrels: new drilling, with high capital intensity and multi-year cycles; improved Niger Delta security against crude theft, pipeline vandalism and illegal refining; new evacuation routes; better terms under the Petroleum Industry Act; licensing rounds; and redevelopment by indigenous operators of divested assets.
- The lowest-cost barrels are in the assets already operated, because the reservoirs are known, the facilities exist, the export routes are connected and the staff are already on payroll.
- THE DOCTRINE
What stands between current output and the achievable potential of existing assets is not geology. It is operational execution.
- Value leaks in routine places. The book's examples: a work order that slips three weeks because its production impact was never quantified; a compressor run degraded for a month because nobody linked its vibration trend to the deferment ledger; field data that takes ten days to reach the decision-maker.
Ask which of those routine leaks is running in your company this week.
6:00 – 12:00
A complex system needs a matching control system
- Nigeria's upstream is several industries sharing a regulator and a coastline: onshore, swamp, shallow water and deepwater assets such as Bonga, Egina and Agbami, with joint-venture assets now substantially in indigenous hands after divestments by Shell, ExxonMobil, Eni and Equinor. Much onshore infrastructure dates from the 1960s through the 1980s.
- NUPRC's oil production status platform puts the physical scale at 323 developed fields, 265 production processing stations and 31 export terminals, linked by more than 5,000 kilometres of pipelines. Every node is a point where performance is gained or lost.
NUPRC published production Liquids, barrels per day July 2025 1.71 million (1.507 million crude plus about 205,000 condensates) January 2026 About 1.63 million February 2026 1.48 million March 2026 About 1.55 million April 2026 1.66 million, crude alone 1.49 million January to February 2026 (recomputed) 1.63 less 1.48 = a fall of about 150,000 Gap from April 2026 to the 3 million ambition (recomputed) 3.00 less 1.66 = 1.34 million - The book reads that volatility as a system still exposed to outages and reliability events that better intelligence could anticipate.
- THE DOCTRINE
Complex systems cannot be managed with fragmented data, delayed reports, manual processes and isolated decision-making. The control system must match the complexity of the system it controls.
A valve failure at one flow station is felt upstream and downstream. So is a slow decision.
12:00 – 18:00
Start with economics, not technology
- Programmes that fail were almost always framed as a technology question rather than a value question.
- The correct executive question is not which AI tool to buy. It is where, specifically, the business is losing production, margin, time, reliability, compliance confidence or operational control, and what each loss is worth per month. The losses cluster: deferment, downtime, well underperformance, delayed maintenance, poor workover prioritisation, spare-parts delays and cost leakage.
- Researchers on the economics of machine intelligence put it simply: AI makes prediction cheap, so the scarce resources become good data and good judgement, and the centre of gravity moves to data quality and decision rights. The International Energy Agency maps the value in energy: subsurface data processing, reservoir simulation, remote operations, predictive maintenance, regulatory compliance, leak detection and automation.
- IN THE FIELD · THE WORKED CASE
- Field. A fictional marginal field operator has a well flowing below potential. It triggers no alarm, which is why it persists. A model that flags it passes the book's test: barrels protected.
IN THE BOARDROOM · THE PARALLEL- Boardroom. The same fictional company is offered a drafting assistant for the executive office. Without a credible path to one of the eight outcomes below, it is not yet strategic.
- The book's test: every proposed AI investment must tie to at least one of barrels protected, downtime avoided, maintenance improved, risks detected earlier, interventions prioritised more accurately, compliance strengthened, costs reduced, or asset visibility improved. If it cannot, it is not yet strategic, however impressive the demonstration.
When a vendor opens with features, answer with your loss and its monthly cost.
18:00 – 22:00
The data problem is conversion, not volume
- Most operators are drowning in data they never use: production reports in spreadsheets whose logic leaves with their author, inspection findings in PDFs no system can query, field notes in WhatsApp, financial impact in systems that close weeks later. Leadership sees it after the window to act has passed.
- A data shortage is solved with more sensors and systems. A conversion problem is solved with architecture, governance and workflow. The book's five operating layers: data ingestion; data governance (named owners, validation, standard definitions of downtime and deferment); AI and analytics; workflow integration, the layer most often missing; and executive visibility of what an event cost and what decision it requires.
- THE DOCTRINE
AI does not create value because it produces an insight. AI creates value when the organisation acts on the insight.
- The data scientists build the engine; the operating model is the drivetrain that puts power to the ground.
Find the layer where your loss disappears from view. That is where conversion fails.
22:00 – 25:00
Write it down, and Section 2
- Before you close the laptop, write one sentence: your loss, its approximate value per month, and the source.
- Section 2 of your plan is The value at stake and our data-to-decision conversion gap.
- The book's charge for this chapter: the next time someone brings you an AI tool, refuse the demo and ask which barrels, which dollars, which month.
A value with no stated source is an opinion. Name the source, even if it is a rough one.
Key Terms
Production deferment
Revenue postponed and, in decline-prone reservoirs, sometimes lost permanently.
Conversion problem
Data exists but does not become intelligence, and intelligence does not become action. Solved with architecture, governance and workflow, not more sensors.
Workflow integration
The fourth layer, most often missing: escalation to a named owner, tasks with deadlines, and follow-up that closes the loop.
Decision rights
Who is entitled and expected to act on a prediction. When prediction is cheap, the book moves the centre of gravity to data quality and decision rights: organisational assets that must be deliberately built.
Self-Check — Before You Submit
1. After an impressive vendor demonstration, which question should a fictional upstream operator's executive committee ask first?
- A. Which model architecture does the tool use?
- B. How many other operators have bought it?
- C. Where are we losing production, margin, time or control, and what is each loss worth per month?
- D. Can it be deployed across every asset at once?
C. Where are we losing production, margin, time or control, and what is each loss worth per month? — Economics decide whether a programme survives its first budget review. Asking about architecture feels rigorous, but it frames a technology question, the documented root cause of failure.
2. Why does the book call barrels from assets already operated the lowest-cost barrels available to Nigeria?
- A. The reservoirs are known, the facilities exist, the export routes are connected and the staff are on payroll
- B. They attract the most favourable fiscal terms under the Petroleum Industry Act
- C. They require no maintenance spending
- D. They are unaffected by evacuation constraints
A. The reservoirs are known, the facilities exist, the export routes are connected and the staff are on payroll — The advantage is what is already in place. Fiscal terms are one of the book's sources of new barrels, not the reason existing assets are cheapest.
3. A fictional downstream, midstream and gas company keeps field notes in WhatsApp and inspection findings in PDFs. Its technology head asks for more sensors. What is the book's diagnosis?
- A. A data shortage, so more sensors are right
- B. A model problem, so a better algorithm is needed
- C. A reporting problem, so a monthly dashboard will fix it
- D. A conversion problem, solved with architecture, governance and workflow
D. A conversion problem, solved with architecture, governance and workflow — The two problems have entirely different solutions. More sensors add volume to data the company already fails to convert; a dashboard alone lacks the workflow that makes anyone act.
Today’s Assignment
State the value at stake in your company's one loss, connect one AI idea to field economics in the book's terms, and name where your data fails to convert into decisions.
- The value at stake. The Section 1 loss per month in barrels, downtime, cost, margin or delay, with the source of each figure. No reliable figure? Say how you will get a baseline.
- One AI idea, tested. Tie one AI application to one of the book's eight outcomes, with a mechanism a finance director would accept.
- The conversion gap. Where the data about your loss lives today, system by system; the weakest of the five layers; its owner by role, or that there is none.
- The honest part. Where your company currently decides on stale or partial information about this loss.
- Four hundred to nine hundred words.
YOUR 90-DAY PLAN
Section 2 — The value at stake and our data-to-decision conversion gap.
How this is marked — the rubric
| States the value at stake per month with a named source, or an honest plan to establish a baseline | 30 |
| Ties one AI idea to one of the book's eight outcomes through a stated mechanism | 25 |
| Traces where the data lives and names the weakest of the five layers with an owner by role | 30 |
| The honest paragraph names a current failing in the student's own company | 15 |
| Total · pass mark 70 | 100 |
Submitting session 2
Write your answer in Word or Google Docs. The first line must read
Write your answer in Word or Google Docs. The first line must read
Session: 2, or the submission is refused. Upload the Word file (.docx), or switch to the Google Doc tab and paste your document’s share link.Marked against the rubric above and returned as a written letter, saved to your progress page. Do not submit until you have passed the self-check above.Capabilities Covered
Field economics before technologyValuing a loss per monthThe eight-outcome testDiagnosing conversion failureThe five operating layers
Part One · The InflectionSESSION 3
The AI Value Chain: From Data to Barrels
Why This Matters
Chapter 1 does not stop at diagnosis. It names where AI pays first in upstream operations: maintenance, project delivery, production, living asset intelligence, and the trust and local capability that hold them up. Each is a link in one chain from data to barrels, and each link fails in its own way. You do not need all of them. You need the one or two that attack the loss you have already named, in an order your company can execute. Today you map two candidate uses onto your company and put them through the book's five leadership priorities, which is where enthusiasm meets discipline.
25-Minute Module
0:00 – 5:00
The value chain and where it breaks
- The upstream AI value chain is simple to state and difficult to execute: Data → Intelligence → Decision → Action → Barrel Impact. It is only as strong as its weakest link.
Link How the book says it fails Data Incomplete maintenance histories, unvalidated allocations, inconsistent sensor tags: weak intelligence that is hard to detect Intelligence Not trusted, because engineers were never involved or early errors were never explained Decision No named owner with a deadline. Insights without owners are merely interesting. Action Not measured against a baseline, so the programme cannot defend its budget and dies regardless of merit - THE DOCTRINE
AI transformation must be led as an operating-model change, not an IT project. IT projects deliver systems. Operating-model changes deliver new behaviours, and barrels come from behaviours.
- Anchor the chain to practical questions: which wells are declining faster than their type curves, which equipment is likely to fail before the next window, which interventions to prioritise this week given limited crews, boats and budget.
The best AI system is the one that helps you make a better field decision before production is lost.
5:00 – 11:00
The two predictive disciplines
- If a leadership team can fund only one AI application in its first year, the book says the evidence strongly favours predictive maintenance. In the book's illustration, a failure with a $200,000 repair cost can destroy several million dollars of value once deferment and logistics are counted, and Nigerian import lead times, customs and FX pressure amplify the multiplier.
- McKinsey cited an offshore operator that applied predictive maintenance across nine platforms in Africa and Latin America: an average 20 percent reduction in downtime and production increases equivalent to more than 500,000 barrels of oil annually. Reactive maintenance tells you what failed; preventive, what to check on a calendar; predictive, what is likely to fail, when, and why it matters commercially.
- Predictive project management protects tomorrow's barrels, because the 2030 ambition is a capital projects ambition. McKinsey found that 98 percent of megaprojects suffer cost overruns of more than 30 percent, and 77 percent are at least 40 percent late. Models can flag slipping work packages weeks before the schedule shows it, and build evidence-based profiles of every contractor.
- THE DOCTRINE
A forecast no one owns is a forecast no one acts on. The weekly project review should debate the forecast, not the history.
Both disciplines turn on one question: does the prediction reach a decision-maker in time?
11:00 – 16:00
Recover barrels, and see the asset as a system
- Production optimisation recovers barrels by making the constraint picture (subsurface, facility, maintenance, evacuation, chemical, commercial) dynamic rather than reconstructed quarterly. In mature and marginal fields the book says many small, unremarkable losses can easily amount to five or ten percent of potential production across a portfolio.
- A digital twin is not a 3D model or a visualisation. It is a living operational representation connecting engineering, production, maintenance, inspection and risk data into one decision-support system.
- IN THE FIELD · THE WORKED CASE
- Field. The book's own chain: a production loss traces to an equipment problem, a delayed work order, a spare part not in stock, a planning process that did not flag it, incomplete equipment data, and finally weak data ownership.
IN THE BOARDROOM · THE PARALLEL- Boardroom. Six links, six departments, one event, each link visible only to its own department. The board sees an equipment failure and the cause repeats. The twin's value is one legible causal chain the board can act on.
- The traditional company asks what happened last month. The intelligent company asks what is changing now and what to do before it affects production.
When a loss is reported as one department's failure, ask for the chain behind it.
16:00 – 21:00
Trust, local capability and the five priorities
- Governance is designed in from the start: a named owner for every model, human approval calibrated to consequence, audit trails, data lineage, cybersecurity and controlled change management, as for plant modifications. NUPRC's April 2026 directive sits within Net Zero by 2060, Zero Routine Flaring by 2030 and a 60 percent methane reduction target by 2035.
- THE DOCTRINE
Do not trust AI because it is advanced. Trust AI when the data, controls, governance and human accountability around it are strong.
- Local Content 2.0. The first phase, anchored by the Nigerian Oil and Gas Industry Content Development Act of 2010 and driven by NCDMB, asked what share of physical work Nigerians do. The next asks what share of the intelligence layer Nigerian companies own. A field operations excellence platform must prioritise backlogs by barrels at risk, not age: dashboards alone change nothing.
The five leadership priorities What the book demands 1. Production value, not technology Quantify each loss; let that ranking, not a vendor's roadmap, set the sequence 2. Data foundation around priority problems Ownership, validation and definitions for the chosen use cases 3. Connect AI to workflow Insight to action, a named owner, measured follow-up 4. Develop local capability No vendor-controlled black box 5. Measure impact in operational terms Barrels, downtime, schedule, cost; never models deployed or pilots completed
If a model's output does not land in someone's working day with a deadline, the model is decoration.
21:00 – 25:00
Write it down, Section 3, and Pop Quiz 1
- The book's promise is precise: the next barrel may come from one avoidable shutdown prevented, one underperforming well found a quarter earlier, or one project reaching first oil on schedule because slippage was forecast six months earlier.
- Before you close the laptop, write down your two candidate uses and the weakest link in each. Section 3 of your plan is Two candidate uses on the value chain, and the leadership priorities that apply.
- Pop Quiz 1 opens now: five questions, closed book, on Sessions 1 to 3. Sit it before Session 4.
Two candidates, honestly tested, are worth more than ten on a slide.
Key Terms
AI value chain
Data → Intelligence → Decision → Action → Barrel Impact. Each link can fail; the chain is as strong as its weakest link.
Remaining useful life
A model's estimate of how long equipment can run before failure. It improves as failure histories accumulate.
Digital twin
A living operational representation of an asset that connects its data into one decision-support system. Not a 3D model.
Local Content 2.0
The share of the intelligence layer (data, models, platforms, analytical capability) owned, built and governed by Nigerian companies.
Self-Check — Before You Submit
1. A fictional marginal field operator with ageing equipment and a slow spares pipeline can fund one AI application this year. Which does the book's evidence favour?
- A. An executive drafting assistant
- B. Predictive maintenance
- C. A full enterprise digital twin
- D. Autonomous production setpoint control
B. Predictive maintenance — Unexpected failure cascades into deferment, emergency procurement and logistics delay. The full twin is the tempting ambitious answer, but it is not the fastest route to value, and setpoint control sits high on the Autonomy Ladder.
2. Which figures does the book attribute to McKinsey's offshore predictive maintenance case?
- A. Twelve offshore platforms worldwide, 30 percent less downtime
- B. Nine platforms in Nigeria, 500,000 barrels per day recovered
- C. Nine platforms in the North Sea, 20 percent lower maintenance cost
- D. Nine platforms in Africa and Latin America, 20 percent less downtime, more than 500,000 barrels annually
D. Nine platforms in Africa and Latin America, 20 percent less downtime, more than 500,000 barrels annually — The case is African and Latin American offshore production. The Nigerian option misplaces the platforms and turns barrels annually into barrels per day.
3. A fictional EPC and construction company's weekly project review spends the hour reconciling last month's progress claims. What does the book prescribe?
- A. Debate the forecast, not the history: which packages are drifting and what to do this week
- B. Move the review to monthly
- C. Ask contractors to submit claims earlier
- D. Add detail to the historical progress report
A. Debate the forecast, not the history: which packages are drifting and what to do this week — The book says that one change in project governance rhythm is worth more than any algorithm. Earlier claims sound helpful but still manage by lagging indicators.
Today’s Assignment
Map two candidate uses from Chapter 1's value chain onto your company and apply the book's five leadership priorities to each.
- Two candidates from predictive maintenance, predictive project management, production optimisation, digital twins, AI-assisted compliance or a field operations excellence platform. At least one must attack your Section 1 loss.
- The chain, twice. For each, what Data, Intelligence, Decision, Action and Barrel Impact would be in your company, and the weakest link today.
- The five priorities. A short answer per priority for each candidate. Where your company fails one, say so.
- The choice. Which goes first, why, and who owns the decision by role.
- The honest part. Which priority your company has most often skipped in past technology projects. Four hundred to nine hundred words.
YOUR 90-DAY PLAN
Section 3 — Two candidate uses on the value chain, and the leadership priorities that apply.
How this is marked — the rubric
| Maps both candidates through all five links in the company's own terms | 30 |
| Applies all five priorities to each candidate, stating failures plainly | 30 |
| Chooses one to go first, tied to the Section 1 loss, with an owner by role | 25 |
| The honest paragraph names a priority the student's own company has skipped | 15 |
| Total · pass mark 70 | 100 |
Submitting session 3
Write your answer in Word or Google Docs. The first line must read
Write your answer in Word or Google Docs. The first line must read
Session: 3, or the submission is refused. Upload the Word file (.docx), or switch to the Google Doc tab and paste your document’s share link.Marked against the rubric above and returned as a written letter, saved to your progress page. Do not submit until you have passed the self-check above.ASSESSMENT NOW OPEN
Pop Quiz 1 — The Thesis and the Value Chain opens now. Sit it before you start the next session — it is how you find out whether the doctrine landed, and the practice papers may be resat as often as you like.
Sit Pop Quiz 1 →Closed book, no notes, and Coach Ritchie switches himself off while you sit it.Capabilities Covered
The AI value chainPredictive maintenance economicsPredictive project controlLiving asset intelligenceThe five leadership priorities
Part One · The InflectionSESSION 4
The Intelligent Energy Company
Why This Matters
An operator can deploy a predictive maintenance model, a digital twin and a surveillance system and still underperform, because those capabilities sit in functions that do not share data, decisions or accountability. Chapter 2 makes the enterprise, not the asset, the unit of competition. That is uncomfortable: the problem is not something you can buy, and you are part of it. Today you trace your Section 1 loss across your company's functions, using the book's own list, to see where the operating model must change.
25-Minute Module
0:00 – 5:00
An operating-model cycle, not a technology cycle
- The industry is not entering a normal technology cycle; it is entering an operating-model cycle. The question is not whether AI becomes more powerful, but whether the enterprise is redesigned to use it.
- Chapter 2 cites Reuters for the NNPC ambitions you met in Session 1, and NUPRC for an operating base linked to export terminals through 5,284 kilometres of pipelines: a distributed, interdependent, high-risk system that needs an operating model acting as one.
- THE DOCTRINE
A barrel lost to a slow decision, a siloed handoff or an unowned process is just as real as a barrel never drilled.
- The intelligent energy company is not the one with the most AI tools. It is the one rebuilt so that intelligence flows from the field to the boardroom and back into action. A redesign, not a procurement.
If your AI plan is a list of purchases, it is not yet a plan.
5:00 – 10:00
The core economic problem
- Most companies have more data and tools than they can use. The operating model wastes them: data scattered and unowned, decisions slow and siloed, work flowing through manual handoffs, accountability blurred. The value leaks in the spaces between functions.
- The symptoms: siloed departments, weak data ownership, manual reporting, slow approvals and no enterprise-level AI governance. You cannot buy your way out of a siloed organisation.
- A company with average tools and an excellent operating model beats a company with excellent tools and an average one, because the operating model decides whether intelligence becomes action.
- Resist two temptations: more tools as a substitute for redesign, and endless redesign without shipping value.
- THE DOCTRINE
The question is not how much AI the company owns. It is how intelligently the company operates.
Look for your loss in the handoffs, not inside any one department.
10:00 – 15:00
What AI actually changes: six capabilities
- AI makes it possible to run an energy enterprise as one real-time system rather than a federation of siloed functions. Six capabilities interlock to do it.
Capability What it does Enterprise AI architecture The connected backbone: common data, shared services, consistent security. Without it each new tool deepens the silos. Agentic workflows with controls Agents that execute steps, introduced last, with human approval, audit logging and strict boundaries. Decision intelligence platforms One trusted basis for action, connecting each insight to a decision, an owner and a tracked outcome. The antidote to dashboard sprawl. Process mining Uses the digital footprints of work to show how processes truly run, replacing opinion about the company with evidence. Knowledge management Captures expertise that otherwise walks out at every retirement and resignation. Executive command centres One real-time view of production, risk, compliance, decisions and exceptions, from which leadership runs the rhythm. - Read the agentic row through the Autonomy Ladder from Session 1: agents that execute steps sit on its highest rung. With governance, agents remove manual drag; without it, they industrialise risk.
- The goal is not to remove human leadership. It is to amplify human judgement with enterprise-wide intelligence, speed and visibility.
Ask which of the six your company has, and which it has bought twice in different departments.
15:00 – 21:00
One trusted data layer, then function by function
- If finance, operations and commercial each keep their own version of the truth, no platform can reconcile them. A deferred barrel, a cost and a production number must mean the same thing everywhere, or the redesign produces faster disagreement rather than faster decisions. The same traceability serves NUPRC's move toward measurement-based reporting.
- Leadership teams change most: from late, backward-looking reports to steering from an enterprise-wide view. A leadership team that reverts to anecdote unmakes the redesign.
Function From To Field operations A source of monthly updates The live edge, consuming and generating intelligence Commercial Periodic analysis on a separate spreadsheet Continuous decisions on the same production data Finance Explaining the past In the room when the decision is made HSE Reactive and paper-based Leading indicators; traceable reporting by design Procurement An administrative bottleneck A delay seen by everyone as a production threat Maintenance A department that fixes things A node that triggers a supplied, coordinated response Business development Instinct Evidence-based cases to partners and investors Regulatory reporting A periodic scramble A continuous, defensible output of the operating model - IN THE FIELD · THE WORKED CASE
- Field. At a fictional upstream operator, a model predicts an export pump failure. Maintenance raises a job; procurement treats the spare as routine administration. The alert is correct and the barrels are still lost.
IN THE BOARDROOM · THE PARALLEL- Boardroom. Finance reconciles the deferment a month later; the committee sees a maintenance failure. Nobody owns the outcome spanning maintenance, procurement and production, where the book says most real losses live.
- THE DOCTRINE
The board does not need nine better departments. It needs one connected enterprise: integration, shared truth and accountability across boundaries.
Draw your loss as it crosses functions. Each crossing is a place to name an owner.
21:00 – 25:00
Write it down, and Section 4
- Before you close the laptop, list the functions your Section 1 loss passes through, in order, and mark the handoff where it most often stalls.
- Section 4 of your plan is How our operating model changes, function by function.
- The book's charge is personal: ask when your leadership team last decided from one shared view instead of nine separate reports. If the answer embarrasses you, that is where the work begins, with you, this quarter.
You cannot delegate this to IT. Write the leadership row yourself.
Key Terms
Operating model
How an enterprise decides, works and holds itself accountable. The book's unit of redesign.
Decision intelligence platform
One trusted view that ties each insight to a decision, an owner and a tracked outcome.
Process mining
Using the digital footprints of work to reveal how processes actually run, so redesign targets real bottlenecks rather than opinion.
Cross-functional accountability
Explicit ownership of outcomes that span functions, where the book says most real losses live.
Self-Check — Before You Submit
1. At a fictional downstream, midstream and gas company, finance, operations and commercial report different volumes for the same month. The sponsor proposes a decision intelligence platform to fix it. What does the book advise?
- A. Buy the platform, since it will reconcile the numbers automatically
- B. Let each function keep its own figure and compare them monthly
- C. First establish shared ownership, standard definitions and validation across functions
- D. Commission a full enterprise data lake before any decision
C. First establish shared ownership, standard definitions and validation across functions — No platform can reconcile separate versions of the truth; without shared definitions you get faster disagreement. The data lake is the opposite trap: the book wants a trusted enough foundation, not perfection first.
2. In what order does Chapter 2 say the intelligent company's capabilities are built?
- A. Foundations and governance first, integration and decision intelligence next, agentic automation last
- B. Agentic automation first to show quick wins, then governance
- C. Command centre first, then data foundation
- D. All six capabilities at once, across the enterprise
A. Foundations and governance first, integration and decision intelligence next, agentic automation last — The company is built in sequence, not switched on. Agents for quick wins is the tempting answer; it is autonomy granted faster than it is governed.
3. What role does process mining play in operating-model redesign?
- A. It automates approvals so that handoffs disappear
- B. It replaces the need for a data foundation
- C. It is a reporting tool for the board
- D. It is the diagnostic: it shows how work truly flows, replacing opinion with evidence
D. It is the diagnostic: it shows how work truly flows, replacing opinion with evidence — You cannot redesign a process you cannot see. Automating approvals answers a different question, and the book introduces agentic automation last, once trust and control are in place.
Today’s Assignment
Describe how your company's operating model changes, function by function, using Chapter 2's own functions.
- Leadership first. How your leadership team decides today, from which reports, how late, and what changes.
- The other functions. For each of the book's eight others that applies to your business, write today and the change. If one does not apply to your company type, say why in a line.
- The worst handoff. The one crossing where your Section 1 loss most often stalls, and a single owner of the cross-functional outcome named by role.
- The data test. One number your functions currently define differently.
- The honest part. Where your company has bought a tool to avoid a redesign. Four hundred to nine hundred words.
YOUR 90-DAY PLAN
Section 4 — How our operating model changes, function by function.
How this is marked — the rubric
| Describes the leadership team's current and changed way of deciding in concrete terms | 20 |
| Covers each relevant function with today and the change, and justifies any omission | 30 |
| Names the worst cross-functional handoff and one owner of the outcome by role, not a department | 25 |
| Identifies one number defined differently across functions | 10 |
| The honest paragraph names a case where the student's own company bought a tool instead of redesigning | 15 |
| Total · pass mark 70 | 100 |
Submitting session 4
Write your answer in Word or Google Docs. The first line must read
Write your answer in Word or Google Docs. The first line must read
Session: 4, or the submission is refused. Upload the Word file (.docx), or switch to the Google Doc tab and paste your document’s share link.Marked against the rubric above and returned as a written letter, saved to your progress page. Do not submit until you have passed the self-check above.Capabilities Covered
Operating-model thinkingThe six enterprise capabilitiesShared data definitionsFunction-by-function redesignCross-functional ownership
Part One · The InflectionSESSION 5
The Value Map, the Rhythm and the Board's Numbers
Why This Matters
A redesign that cannot show its value will be quietly defunded. One with value but no rhythm drifts back into silos. One with a rhythm but no baseline proves nothing to a board. The second half of Chapter 2 closes those gaps: a commercial value map, a disciplined 90-day pilot and operating rhythm, and board metrics tied to baselines. Today you write the version that fits your company, and it must survive a hard question. Where you have no baseline yet, say so. That admission is the start of measurement, not a weakness.
25-Minute Module
0:00 – 5:00
The commercial value map
- Value is mapped at the enterprise level, because the returns come from connections no single function captures. The book's map has four parts:
Part of the map The question it asks The cost of silos What was deferred because a decision waited on another function? What did a procurement delay cost production? What margin was lost because commercial and operations used different numbers? Speed How long does the enterprise take to go from signal to decision to action? Durability and trust Is the company credible to regulators, investors and talent because it operates with evidence, control and accountability? Compounding Does better data enable faster decisions, better results and more capability, which improves the data again? - THE DOCTRINE
Decision cycle time is the master variable of the intelligent company. Every reduction in it compounds across every function.
- Cross-functional losses are usually larger than any one function's inefficiency. Point solutions give one-time gains; the operating model is a flywheel.
Start your map with one decision that waited on another function, and what the wait cost.
5:00 – 10:00
The 90-day pilot, the roadmap and scaling
- The pilot targets one high-value cross-functional decision. Days 1–30: map the process with process mining, quantify its friction, name the owners, and write a charter of scope, baseline, decision, data and accountability. Days 31–60: connect the data, stand up the decision view, define the workflow and its controls. Days 61–90: run it live and track decisions to outcomes. Then continue, expand, adjust or stop.
- The book warns where redesigns most often fail: the integrated capability is built, the silos quietly reassert themselves, and the workflow reverts.
- THE DOCTRINE
A pilot succeeds only when it proves a repeatable pattern of integration. Anything less is a demonstration, not a redesign.
- The roadmap's five principles: value case before technology, validated with finance; map and validate the data; design the decision workflow before the dashboard; pilot on a high-value use case; scale only after adoption is proven.
- Scaling needs standardisation, sequencing (foundation and governance, then decision intelligence and integration, then agentic automation; the book's phasing figure is illustrative) and ownership by the executive team, with a named senior owner per capability. Delegated to IT, it produces tools.
Decide the stop criteria before day one, while nobody is attached to the pilot.
10:00 – 15:00
The executive operating rhythm
- The operating model becomes real in leadership's rhythm. The command centre is the instrument; the book's mock-up of one is illustrative.
Rhythm Runs on The book's questions Daily Exceptions across the enterprise What changed since yesterday? Which cross-functional risks are rising? Which decisions are stuck between functions? Which actions are overdue? Weekly Integration and accountability Which cross-functional decisions improved an outcome? Which silo frictions caused delay? Who owns which integrated process? Monthly Performance, capability and strategy execution How are KPIs trending against baseline? Where is the model still siloed? Which strategic decisions should be accelerated? - IN THE FIELD · THE WORKED CASE
- Field. At a fictional fabrication yard, the daily review shows a materials delivery stuck between procurement and the yard planner. Raised as an exception with an owner, it is cleared before it costs a milestone.
IN THE BOARDROOM · THE PARALLEL- Boardroom. The same fictional yard's monthly review asks why that friction recurs and whether cross-functional accountability is improving against baseline. The same loop, at a different speed.
- Sponsorship is leadership of the redesign, not endorsement. The Field Operations Excellence Lens sets the target: a company operationally disciplined, AI-enabled, locally capable and commercially accountable.
If the weekly meeting still reads departmental reports in turn, the rhythm has not changed.
15:00 – 21:00
Board metrics, inaction and trust
- The board should not measure tools deployed, pilots run or dashboards built. Those are activity metrics. The book's eight enterprise KPIs: decision cycle time; workflow automation rate; AI adoption quality; operating margin improvement; production reliability; reporting accuracy; cross-functional accountability; strategy execution velocity. The book labels its figure plotting them an illustrative index.
- THE DOCTRINE
Each KPI should be tied to a baseline. If the company cannot define today's figure, it cannot credibly claim improvement.
- Add integration and capability checks: are the functions sharing one trusted view, are silos reforming, is indigenous capability deepening, is governance keeping pace? The final metric is learning velocity.
- The risk of inaction is the company that digitises but does not redesign. It compounds silently: operational, financial, compliance and talent consequences.
- Trust. The book cites the NIST AI Risk Management Framework and ISO/IEC 42001 for governing AI, and NIST SP 800-82 for securing operational technology. A compromised intelligent enterprise fails in more places at once than a siloed one. Leadership must say which decisions stay with people, always. Local Content 2.0 now means owning the operating model itself.
Choose metrics you can baseline this month, not the ones that look best in a board pack.
21:00 – 25:00
Write it down, Section 5, and Practice Exam 1
- Before you close the laptop, write the three board metrics you will propose and, beside each, today's figure or the words no baseline yet.
- Section 5 of your plan is Our commercial value map, executive operating rhythm and board metrics with baselines.
- Week 1 closes here. Practice Exam 1 opens now: fifteen questions, closed book, covering Sessions 1 to 5. Sit it before Week 2.
- The chapter's leadership conclusion: AI can help, but leadership must define the operating model, sequence the capabilities, govern the risk, develop the people and measure the results.
A metric without a baseline is a hope. Write the baseline, or write that you do not have one.
Key Terms
Decision cycle time
How long the company takes to move from signal to decision to action. The book's master metric of the intelligent enterprise.
Activity metric
A count of effort, such as tools deployed, pilots run or dashboards built. The book says boards should measure outcomes instead.
AI adoption quality
Whether AI is used well: adopted into real decisions, trusted by its users and producing better outcomes, not merely deployed.
Baseline
Today's measured value of a metric, fixed before a change, so that any improvement can be credibly claimed and audited.
Self-Check — Before You Submit
1. A fictional EPC and construction company's AI board report lists pilots completed, dashboards launched and users trained. What should the board ask for instead?
- A. A larger number of pilots next quarter
- B. Enterprise outcome KPIs, such as decision cycle time, each against a baseline
- C. A demonstration of the most advanced dashboard
- D. The vendor's adoption statistics for comparable companies
B. Enterprise outcome KPIs, such as decision cycle time, each against a baseline — The report lists activity metrics, which the book says the board should not use. More pilots is the tempting reply, but it measures effort, and without a baseline no improvement can be credibly claimed.
2. In month three of a cross-functional pilot at a fictional upstream operator, the integrated view works, but managers have gone back to departmental reports. What is happening?
- A. The pilot has succeeded and should be scaled
- B. The technology is faulty and should be replaced
- C. The pilot ran too long and should be shortened
- D. The silos are reasserting themselves, the most common failure in enterprise redesign
D. The silos are reasserting themselves, the most common failure in enterprise redesign — Redesigns most often fail this way. Scaling now would multiply fragility; a pilot succeeds only when it proves a repeatable pattern of integration.
3. Which references does Chapter 2 name for governing AI across the enterprise and for securing operational technology?
- A. The NIST AI Risk Management Framework and ISO/IEC 42001 for AI; NIST SP 800-82 for operational technology
- B. NUPRC's GHG reporting guidance for both
- C. The Nigerian Oil and Gas Industry Content Development Act for AI; NUPRC templates for operational technology
- D. The Petroleum Industry Act for both
A. The NIST AI Risk Management Framework and ISO/IEC 42001 for AI; NIST SP 800-82 for operational technology — The book cites exactly these three. NUPRC and the local content Act appear in the book, but for reporting and local content, not as AI governance or operational technology security frameworks.
Today’s Assignment
Draft your company's commercial value map entry, its executive operating rhythm, and three board-level metrics with baselines.
- The value map. For your Section 1 loss, write the four parts: the cost of silos, speed, durability and trust, and compounding. State figures with sources, or not known.
- The rhythm. Daily, weekly and monthly: who chairs, which view they use, and the questions asked. Describe today's rhythm first, then the changed one.
- Three board metrics from the book's eight. For each: today's baseline and its source, or how and by when it will be measured, and the owner by role.
- The honest part. Which activity metrics your company currently reports to its board as if they were outcomes.
- Four hundred to nine hundred words.
YOUR 90-DAY PLAN
Section 5 — Our commercial value map, executive operating rhythm and board metrics with baselines.
How this is marked — the rubric
| Completes all four parts of the value map for the company's own loss, with sources or unknowns marked | 30 |
| Sets out daily, weekly and monthly rhythms with a chair by role, a view and the questions asked | 25 |
| Chooses three of the book's eight KPIs, each with a baseline or a dated plan to establish one and an owner by role | 30 |
| The honest paragraph names activity metrics the student's own company reports as outcomes | 15 |
| Total · pass mark 70 | 100 |
Submitting session 5
Write your answer in Word or Google Docs. The first line must read
Write your answer in Word or Google Docs. The first line must read
Session: 5, or the submission is refused. Upload the Word file (.docx), or switch to the Google Doc tab and paste your document’s share link.Marked against the rubric above and returned as a written letter, saved to your progress page. Do not submit until you have passed the self-check above.ASSESSMENT NOW OPEN
Practice Exam 1 — The Inflection opens now. Sit it before you start the next session — it is how you find out whether the doctrine landed, and the practice papers may be resat as often as you like.
Sit Practice Exam 1 →Closed book, no notes, and Coach Ritchie switches himself off while you sit it.Capabilities Covered
Enterprise value mappingDisciplined 90-day pilotsExecutive operating rhythmOutcome metrics with baselinesScaling without losing control