10-Session Programme · 25 Minutes Per Day · Free · For leaders in Nigeria’s oil and gas sector
WEEK 2 OF 2 · THE SERVICE COMPANY, TRUST AND DELIVERY · SESSIONS 6–10 · WORK AT YOUR OWN PACE
Part Three · The Intelligent Service CompanySESSION 6
Tenders, Quotes and Mobilisation
Why This Matters
Week 2 follows the evidence into the service company, where a recommendation becomes a tender, a price, a mobilised crew and, eventually, cash. If your company is a contractor, this is where the book places eligibility control, scope-to-cost traceability, margin definitions and readiness, all before a single job starts. If your company is an operator, these are the decisions your contractors make because of the tender, scope and prerequisites you hand them. Either way, I would ask you to read today as a question about control: who checks the eligibility, the arithmetic and the readiness before anyone commits money or people.
25-Minute Module
0:00 – 6:00
Opportunities: eligibility before excitement
- Chapter 12's test: an opportunity is useful when the organisation can recognise it, determine eligibility and respond within time. Deterministic collection of permitted portal changes, notices, permit renewals and certification expiries preserves the source, retrieval time and original notice; AI classifies relevance and extracts candidate requirements.
- THE DOCTRINE
Silence is ambiguous: it can mean no new notice or a broken collection process. A leadership team cannot rely on an apparent absence of tenders unless monitoring health is known.
- Hold a capability and eligibility register separating company certificates, individual credentials, equipment certifications and project-specific permits, each with issuing authority, scope, expiry, verification date and owner. An expired or out-of-scope credential does not become valid because a summary says the organisation is qualified.
- Qualification examines technical scope, capacity, commercial terms, data rights, HSE obligations and working-capital exposure, with documented bid or no-bid reasons. A model trained on past wins can reproduce a bias toward familiar clients.
- Calculate deadlines deterministically and link addenda to the tender version. Each compliance matrix row traces to a source clause with an owner, evidence and status. Never let the system fabricate a reference project, qualification or equipment specification.
- Win rate alone rewards easy bids; examine qualified win rate, contribution, cash exposure and delivery performance.
When the digest is quiet, ask first whether the collector ran, before concluding the market is quiet.
6:00 – 13:00
The quote: controlled quantities and honest definitions
- Chapter 13: AI is useful for detecting omissions and contradictions, but arithmetic and pricing rules remain deterministic and reviewable. A language model should not silently calculate a bill of quantities, invent a unit rate or determine an unapproved margin.
- Map every scope element to a cost item or an explicit exclusion, from preparation and mobilisation to standby and close-out.
- Units carry commercial meaning: a complete system rate is not a rate for one coat; crew days are not equipment days. The estimator, not the AI, establishes the definition. The relationship is expected cost = quantity × consumption per unit × unit cost, adjusted for stated recovery or loss assumptions where justified.
- Treat uncertainty before margin: separate known cost, estimated cost, scenario allowance and commercial margin. A contingency does not replace clarifying an ambiguous scope.
- THE DOCTRINE
A 25 percent markup on cost produces a 20 percent gross margin. Encode the definition in the pricing model and show it in the approval record.
Appendix C Model 5 (illustrative, generic currency units) Base case Consumption rises Quantity × consumable demand 2,000 m² × 30 kg/m² = 60,000 kg (60 t) 2,000 m² × 40 kg/m² = 80,000 kg Consumables at 400 per kg 24 million 32 million Other direct and allocated costs 16 million 16 million Total estimated cost 40 million 48 million Price for 20% gross margin: 40 ÷ 0.8 50 million Held at 50 million Price for 20% markup on cost instead 48 million, margin 8 ÷ 48 = 16.67% — Gross margin at 50 million 20% 2 ÷ 50 = 4% - Clarifying consumption and method can matter more than refining a forecast to another decimal place. None of these rates is a Nigerian quotation.
If two people in your commercial team would give different answers to what our margin is on this job, the definition is not yet in the pricing model.
13:00 – 19:00
Independent review, then readiness rather than movement
- The review sequence is scope extraction, estimator build, AI completeness review, technical review, commercial approval and controlled issue. The AI review returns exceptions with clause references; it does not authorise the quotation or certify technical feasibility.
- Chapter 14: mobilising before the site is ready creates cost and exposure. Readiness means scope approval, access, permits, competence, equipment condition, materials and client prerequisites. A missing prerequisite is a constraint, not a problem that will somehow be solved in time.
- A mixed-integer scheduling model is, in plain terms, a mathematical planning aid that chooses job assignments and start times within the constraints you give it. Its output is only as valid as its constraints and assumptions. Utilisation means accepted work: track availability, mobilisation, operation, standby, maintenance and quarantine separately.
- THE OPERATOR · THE WORKED CASE
- Operator. A fictional operator issues an addendum changing a scope element while an access approval remains pending on its side. Unless the addendum is linked and the prerequisite visible, its contractor prices and mobilises against an obsolete version, and the slipped approval becomes the contractor's standby.
THE SERVICE COMPANY · THE PARALLEL- Service company. The fictional contractor's readiness review freezes the affected commitment and re-evaluates, without automatically substituting people or equipment. The action owner confirms the revised plan respects client acceptance, technical requirements and commercial terms.
- THE DOCTRINE
If mobilisation repeatedly slips because documentation is late, improve document readiness. Predictive scheduling should help change recurring causes, not merely forecast dysfunction more accurately.
Plan with a baseline and a small set of credible scenarios, and ask which jobs become infeasible. The book finds scenarios often more interpretable than unverified distributions.
19:00 – 25:00
The playbooks, your section and the assignment
Appendix D playbook Who verifies What to measure Daily tender and compliance digest Business development; operations Qualified opportunities, missed notices found later, unsupported classifications, owner response Quotation completeness review Technical and commercial reviewers Omissions found after issue, revision causes, contribution lost to unapproved scope Mobilisation readiness review Operations manager; commercial owner checks funding Avoidable standby, late readiness, resource conflicts, accepted productive utilisation - No automated reviewer can guarantee completeness, so independent review is proportionate to consequence.
- The leadership charges together: require the opportunity process to report its own coverage and failures; approve a price only when the organisation can explain what it will deliver and how it will fund delivery; review readiness and resource conflicts before approving mobilisation.
- Write down where your tender-to-mobilisation flow decides on unchecked evidence. That starts Section 6 of your programme: Our tender, quote and mobilisation decisions and the controls they need. Then write the assignment.
A contingency is not a substitute for clarifying an ambiguous scope. Look upstream of the margin.
Key Terms
Capability and eligibility register
A controlled record of certificates, credentials, equipment certifications and permits, each with issuer, scope, expiry, verification date and owner.
Compliance matrix
A table tracing each tender requirement to its source clause, with owner, evidence and status. Gaps stay visible.
Markup versus gross margin
Markup is applied to cost; gross margin is measured on selling price. A 25 percent markup on cost gives a 20 percent gross margin.
Accepted productive utilisation
Equipment and crew time spent on work the client accepted, as distinct from time merely allocated on a schedule.
Self-Check — Before You Submit
1. A fictional contractor's tender digest has shown no new opportunities for three days, and the collector log shows failed page fetches. What should the business development lead do?
- A. Report a quiet market to the managing director
- B. Ask AI to summarise the last successful notices again
- C. Treat the silence as ambiguous and escalate the missed sources
- D. Lower the relevance threshold so more leads appear
C. Treat the silence as ambiguous and escalate the missed sources — Chapter 12 says silence can mean a broken collection process, and leadership cannot rely on an apparent absence of tenders unless monitoring health is known. Lowering the threshold rewards weak leads and ignores the unchecked sources.
2. A pricing model applies a 25 percent markup on cost. What gross margin on selling price does that produce?
- A. 20 percent
- B. 25 percent
- C. 16.67 percent
- D. 31.25 percent
A. 20 percent — On a cost of 100, a 25 percent markup gives a price of 125, and 25 ÷ 125 = 20 percent, as Chapter 13 states. Answering 25 percent is exactly the confusion the book says the pricing model and approval record must prevent.
3. In Appendix C Model 5, consumable demand rises from 30 to 40 kilograms per square metre while the price stays at 50 million currency units. What happens to gross margin?
- A. It falls from 20 percent to 16.67 percent
- B. It is unchanged because consumables pass through
- C. It falls from 20 percent to 10 percent
- D. It falls from 20 percent to 4 percent
D. It falls from 20 percent to 4 percent — Consumables rise to 32 million and total cost to 48 million, leaving 2 million on 50 million: 4 percent. The 16.67 percent figure belongs to the separate markup comparison. The example is illustrative.
Today’s Assignment
Write Section 6 of your First-Year AI Operating Programme: one tender-to-mobilisation flow in your company, or for an operator one contractor engagement, and the controls its decisions need.
- Trace the flow in your business from opportunity to approved mobilisation. Operators trace one contractor engagement from tender to client prerequisites.
- Say who owns your eligibility records and whether an expired credential could reach a bid unnoticed.
- State how your company defines markup and margin, whether every scope element maps to a cost item or exclusion, and which assumptions are routinely unresolved at issue.
- List your readiness prerequisites with evidence and owner, and the most common reason capacity is lost.
- Name the one point where intelligence would change a decision, its Autonomy Ladder permission and its link to your Session 1 decision.
- Be honest about what your company currently gets wrong. If your company has no AI in use yet, that is normal: work from the decision in front of you. Write 400 to 900 words.
YOUR 90-DAY PLAN
Section 6 — Our tender, quote and mobilisation decisions and the controls they need.
How this is marked — the rubric
| Traces one complete flow in the student's own company with owners by role at each stage | 25 |
| States the markup and margin definition, scope-to-cost traceability and unresolved assumptions accurately | 25 |
| Lists readiness prerequisites with evidence and owners and names a recurring cause of lost capacity | 20 |
| Names one decision intelligence would change, its Autonomy Ladder permission and the Session 1 link | 20 |
| Gives an honest account of what the company currently gets wrong | 10 |
| Total · pass mark 70 | 100 |
Submitting session 6
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: 6, 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
Monitored opportunity processEligibility controlQuote traceabilityMargin definitionsMobilisation readiness
Part Three · The Intelligent Service CompanySESSION 7
Job Execution, Job Economics and Learning From Reports
Why This Matters
A job is not finished when the crew demobilises. It is finished when the evidence is accepted, the invoice is accepted and the cash is in the bank, and each of those steps can fail after the technical work has gone well. Today follows one job from its plan to collected cash, and into the status reports managers rely on. For a contractor, this is where a profitable quote can become a funding problem. For an operator, it is where your acceptance requirements and approvals decide how quickly your contractor can close. I would ask you to look for the leak in your own company, not the theory.
25-Minute Module
0:00 – 6:00
The job record is part of delivery
- Chapter 15 asks for a controlled plan, an execution record and acceptance criteria on every job. A planned activity is not proof that it occurred, and an internal completion statement is not client acceptance.
- Build the job data schema around the specific service, from authorised scope to as-left condition, and capture evidence where the work occurs, not at invoice time.
- An inspection result keeps its instrument identity, calibration status, method and responsible inspector. AI may flag an outlier; the inspector decides whether it is a defect, a recording error or a legitimate condition.
- THE DOCTRINE
An assistant should never infer an omitted witness approval from the fact that subsequent work continued. Absence of the approval is an exception requiring investigation.
- Nonconformities are learning records: requirement, departure, containment, corrective action and closure evidence. A model should not label a person incompetent because their jobs carry more recorded nonconformities; better reporting may raise recorded events while improving transparency.
Require evidence at the point where work occurs. Do not leave the field team to reconstruct an entire job at invoice time.
6:00 – 12:00
From acceptance to margin
- Map deliverables to the acceptance requirements agreed before mobilisation. Missing signatures, serial numbers or approved variations can delay invoicing even when the work succeeded. Measure first-pass acceptance, rework exposure and time to accepted close-out, alongside safety and quality.
- Chapter 16: a quoted margin can disappear through extra work, standby, consumption, exchange exposure and delayed collection. An estimate-at-completion process uses actual commitments and credible remaining cost, compared with the quoted assumptions, not hidden behind an unexplained profitability score.
State What can go wrong Revenue eligibility A variation is technically completed but commercially unapproved Invoice issue An invoice is issued but rejected for missing documentation Cash collection Cash arrives after significant financing cost - THE DOCTRINE
These states require different decisions and should not be collapsed into one progress percentage.
- Contract intelligence has limits. An assistant can locate clauses and distinguish an extracted clause from an interpretation; it does not determine enforceability. The commercial owner decides, and no agent negotiates terms or waives rights to speed completion.
Ask for three numbers separately: earned, invoiced, collected. If nobody can give you all three, you have found the leak.
12:00 – 18:00
Funding the job, and measuring what was collected
- Build a dated cash schedule for mobilisation, procurement, payroll, transport, execution and collection. The peak cumulative deficit measures funding need: the lowest point the job's running cash balance reaches, like the lowest level a storage tank falls to before the next delivery arrives.
Appendix C Model 6 (illustrative, millions of currency units) Cash movement Cumulative position Start: advance received +20 +20 Period one: spend −30 −10 Period two: spend −15 −25 Period three: receipt +25 0 Period four: receipt +20 +20 - The peak deficit is 25 million, before financing and omitted costs. A profitable job still needs that funding, and a delayed first receipt makes the deficit last longer. Include only confirmed advances: a contractual right to an advance is not cash in the bank. The model does not create liquidity and should not approve an unfunded mobilisation.
- THE OPERATOR · THE WORKED CASE
- Operator. In a fictional case, the operator's witness sign-off and variation approval wait on one engineer. Each week of delay extends its contractor's deficit and leaves the operator's own records incomplete. Your acceptance process is part of your contractor's cash schedule.
THE SERVICE COMPANY · THE PARALLEL- Service company. The fictional contractor runs a range of payment dates, and exchange-rate scenarios where relevant, and finds it is profitable on accrual yet unable to fund the next mobilisation. Working capital can decide whether a bid is feasible at all.
- Measure collected contribution: a consistent management measure, distinct from statutory accounting, with explicit cost boundaries, read with cash collection and financing exposure. For AI investments, staff time saved is not automatically a cash saving.
A job that makes money and cannot be funded is still a job you should not have started on those terms.
18:00 – 22:00
Reports that change decisions
- Chapter 17 concerns Project and Personal Status Reports (PSRs). The discipline begins with a controlled identifier, reporting period, responsible author and submission cut-off. A late report remains late even if AI rewrites it.
- Rules check fields, dates and identifiers; AI checks meaning, such as vague completion evidence, and returns exceptions for the author to verify. Green needs evidence it is within plan; red names the consequence and the decision needed.
- AI can propose coaching questions from observable patterns, but cannot infer attitude, intent or competence from writing style. Avoid automated disciplinary decisions; report-level scores must be explainable and contestable. A mismatch between reported completion and job acceptance triggers a review, not an accusation.
- THE DOCTRINE
Repeatedly coaching individuals to write better reports will not resolve a procurement process that continually blocks delivery.
When the same blocker appears in five people's reports, it is not five reporting problems. It is one management problem.
22:00 – 25:00
Your section and the assignment
- Appendix D closes the loop. Service close-out review: compare actual work with approved scope, verify the evidence, and track time to acceptance and causes of rework. PSR review and coaching: distinguish on-time submission, completeness, factual quality and unresolved action, not one score.
- The leadership charges: treat the job record as part of delivery; ask finance and operations to explain both the forecast margin and the dated funding requirement; keep factual verification, coaching judgment and employment decisions with accountable people.
- Write down where, in your last completed job or contract, evidence, margin or learning leaked. That opens Section 7 of your programme: From plan to collected cash: our execution evidence, job economics and reporting discipline. Then write the assignment.
A digitally complete pack must also be a truthful account of the job. Hold your people, and your contractors, to both.
Key Terms
Estimate at completion
The forecast final cost of a job, built from actual commitments plus credible remaining cost and compared with the quoted assumptions.
Peak cumulative deficit
The deepest point a job's cumulative cash position reaches before receipts recover it; the book's measure of the funding a job needs.
Collected contribution
A consistent management measure of job contribution, with explicit cost boundaries, read alongside cash actually collected and financing exposure.
First-pass acceptance
The share of deliverables or close-out packs the client accepts at first submission, without rework.
Self-Check — Before You Submit
1. In Appendix C Model 6, a job receives an advance of 20 million, spends 30 million then 15 million, and later receives 25 million and 20 million. What is its peak funding need before financing?
- A. 10 million
- B. 25 million
- C. 45 million
- D. 20 million
B. 25 million — The cumulative position runs +20, −10, −25, 0, +20, so the peak deficit is 25 million. 45 million is total spend, ignoring the advance; 10 million is only the position after period one. The figures are illustrative.
2. A fictional contractor's job pack shows a hold point with no witness approval, though the next activity went ahead. The close-out assistant proposes marking it witnessed. What is the right response?
- A. Accept it, because later work implies approval
- B. Delete the hold point from the plan
- C. Have the supervisor sign it now with the original date
- D. Record the missing approval as an exception and investigate
D. Record the missing approval as an exception and investigate — Chapter 15 says an assistant should never infer an omitted witness approval from continued work. Backdating a signature is the most tempting shortcut and the most damaging: it turns a gap into a misrepresentation.
3. A manager's PSR reports a task complete, but the client has not accepted the related job. What does Chapter 17 direct?
- A. Trigger a review, since different periods or definitions may explain the mismatch
- B. Issue a formal warning for inaccurate reporting
- C. Mark the report red and score the author down automatically
- D. Ignore it, because PSRs and job records serve different purposes
A. Trigger a review, since different periods or definitions may explain the mismatch — A mismatch triggers a review rather than an accusation, and the resolution improves the schema. A warning or automated score is tempting but crosses into employment decisions that stay with accountable people.
Today’s Assignment
Write Section 7 of your First-Year AI Operating Programme: one job in your business traced from plan to collected cash, naming where evidence, margin or learning leaks today.
- Choose one recent job in your company; operators may trace one contractor job. Describe its plan, execution record and acceptance criteria as they actually existed.
- Name where evidence was missing or reconstructed late: measurements, calibration, witness approvals, variations, signatures.
- Trace the margin from quote to estimate at completion, and separate revenue eligibility, invoice issue and cash collection. Sketch the peak cumulative deficit if you can; if not, say who could.
- Name one recurring constraint your status reports reveal that leadership has not acted on.
- Name one place AI could help, with its permission, owner and limits, connected to your Session 1 decision.
- Be honest about what your company gets wrong. If your company has no AI in use yet, work from the job in front of you. Write 400 to 900 words.
YOUR 90-DAY PLAN
Section 7 — From plan to collected cash: our execution evidence, job economics and reporting discipline.
How this is marked — the rubric
| Traces one real job in the student's own company from plan through acceptance to collected cash | 25 |
| Separates revenue eligibility, invoice issue and cash collection and addresses the funding requirement | 25 |
| Identifies specific evidence gaps and a recurring reporting constraint with its cause | 20 |
| Names one AI use with permission, owner and limits, linked to the Session 1 decision | 20 |
| Gives an honest account of where the company leaks evidence, margin or learning | 10 |
| Total · pass mark 70 | 100 |
Submitting session 7
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: 7, 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
Evidence-led executionAcceptance and close-outMargin tracingFunding needStatus report discipline
Part Four · Trust and ControlSESSION 8
Safety, Environment, Community and Nigerian Capability
Why This Matters
This session covers the records with the least tolerance for optimism: safety, environment, community commitments, regulatory obligations and Nigerian capability. Each tempts an organisation to let a model's output stand in for evidence: a hazard list for an assessed hazard, an inferred quantity for a measurement, a headcount for capability, a summary for a regulation. The book's position is consistent and, I think, right. AI can help find, compare and flag. It does not protect, certify or confirm. Your job is to keep that boundary visible and give each record an owner.
25-Minute Module
0:00 – 6:00
The boundary between prediction and protection
- Safety applications need particular care because severe events are rare and their consequences large. A model may identify leading indicators, retrieve procedures or flag missing evidence. That does not establish its suitability as a protective function. Independent safety systems and approved procedures keep their engineered roles.
- Think of a trend chart that tells a supervisor where to look, beside the engineered protection that acts on its own. A safety model sits on the trend-chart side of that line.
- The official upstream safety regulations and relevant NUPRC HSEC guidance are starting points for an applicability review. The book's sources are the Nigerian Upstream Petroleum Safety Regulations 2024 on the NUPRC gazetted register [3] and the Petroleum Industry Act 2021 [2]. A compliance process identifies the particular operation, role, instrument and current version. The book's tools do not certify compliance or prescribe operating limits.
- THE DOCTRINE
Ask what happens if the safety model is wrong or unavailable. Preserve independent protection.
- AI can compare a proposed task with approved hazard information, but must name its source and distinguish a candidate hazard from an assessed hazard. A generated list is not evidence that controls are in place.
- For simultaneous operations, stale location or isolation data can make an attractive visualisation unsafe.
Before you approve any safety-related model, ask for the page that says what the site does when it is switched off.
6:00 – 12:00
Environmental and community evidence
- For emissions, leaks or waste records, distinguish direct measurement, inference and estimate; calibration, detection limits, weather and survey coverage matter. A model can prioritise investigation. It should not represent an inferred quantity as a verified measurement for reporting.
- Track an environmental exception through investigation, repair, verification and record closure. A suspected leak removed from a dashboard is not necessarily repaired. The record says whether it was confirmed, rejected or unresolved, and why.
- Community records may hold personal information, disputes and commitments. Sentiment scores do not substitute for engagement or evidence of consent; avoid opaque models that classify communities as hostile or justify surveillance.
- The PIA provides the legal context for host-community arrangements; determine the relevant entity and obligation before building a workflow [2]. Preserve the distinction between statutory arrangements, contract commitments and voluntary actions. Do not invent beneficiaries, completion evidence or community approval.
Record The statuses the book requires you to keep apart Hazard Candidate hazard, assessed hazard Environmental quantity Direct measurement, inference, estimate Environmental exception Confirmed, rejected, unresolved, with the reason Community commitment Statutory arrangement, contract commitment, voluntary action Regulatory instrument Draft, final; publication date, commencement date
The book requires evidence before a status changes. Of any changed status, ask who changed it, and on what.
12:00 – 18:00
Obligations as controlled records
- Chapter 19: a regulatory assistant works from verified official instruments and authorised internal interpretations. Track title, issuer, status, effective date where confirmed, version, applicability and owner. A draft must not be presented as final, and a publication date may differ from commencement.
- The book practises this itself: the National Data Repository Regulation on the NUPRC draft register is not treated as proof of a final operative instrument [13].
- Translate obligations into tasks with evidence and review dates, linked to the source. The model may propose an obligation; the compliance or legal specialist confirms meaning and applicability.
- Report which sources were checked; a failed fetch is an exception, and humans review the source register periodically. Do not promise that AI will ensure nothing is missed; establish measurable coverage and escalation.
- THE DOCTRINE
A compliance register's value comes from evidence of verification and action, rather than from the number of entries it contains.
- THE OPERATOR · THE WORKED CASE
- Operator. A fictional operator holds the obligation register for its asset. When a contractor's scope touches an obligation, it links the task to the source instrument and names the owner, rather than passing on a general summary.
THE SERVICE COMPANY · THE PARALLEL- Service company. The fictional contractor holds the evidence that the obligation was met on the job: credentials, competence records, equipment certifications. The two must agree on identifiers, versions and who verifies what.
Ask to see the exceptions the register has produced. A failed page fetch must produce one.
18:00 – 22:00
Nigerian capability you can demonstrate
- The Nigerian Oil and Gas Industry Content Development Act establishes a statutory framework for Nigerian content and the role of the monitoring board [4]. AI can connect requirements with personnel, procurement and training evidence, but the requirements come from the authoritative instrument and approved project arrangements.
- THE DOCTRINE
Measure capability as demonstrated work, not simply local headcount or an attendance certificate.
- Record the task a person can perform, the assessment method, supervision conditions and evidence. Link training to opportunities for practice. A model can suggest learning gaps; a competent assessor verifies competence.
- Retain knowledge through rights and design: a governed library of procedures, lessons and model specifications, with contractual rights for access, use and portability.
- Ask vendors for limitations, test evidence, incident response, data-use terms and exit arrangements. A demonstration on a curated dataset does not establish performance on local equipment, documents or connectivity.
If the knowledge leaves when the contract ends, you rented it. Decide that deliberately, not by default.
22:00 – 25:00
Your section, the assignment and Pop Quiz 2
- The leadership charges: require evidence before changing the status of a hazard, environmental issue or community commitment; require verified applicability for compliance and demonstrated capability for local development, each with an owner.
- Write down the one record in your company whose status could change without evidence. That begins Section 8: Our protection boundary, our obligations as controlled records, and measurable Nigerian capability. Then write the assignment.
- Pop Quiz 2 opens now: five questions on Sessions 6 to 8, closed book. Sit it before you start Session 9.
Protection, obligation and capability have one thing in common: each needs evidence that someone verified it, not a model that said so.
Key Terms
Independent protection
Safety systems and approved procedures that keep their engineered role whatever a model says, and still work if it is wrong or unavailable.
Candidate hazard
A hazard proposed by a system or search, not yet assessed by the authorised permit and hazard process. It is not evidence that controls are in place.
Controlled obligation record
An obligation held with title, issuer, status, confirmed effective date, version, applicability and owner, linked to tasks and evidence.
Demonstrated capability
Evidence that a person can perform a defined task, with assessment method, supervision conditions and assessor recorded.
Self-Check — Before You Submit
1. A fictional operator's leak-detection dashboard no longer shows a suspected leak flagged last week. The asset manager wants to report it closed. What comes first?
- A. Report it closed, since the alert has cleared
- B. Ask the model to re-score the area
- C. Establish whether it was confirmed, rejected or unresolved, and verify repair
- D. Remove the sensor from monitoring
C. Establish whether it was confirmed, rejected or unresolved, and verify repair — Chapter 18 says a suspected leak removed from a dashboard is not necessarily repaired; the exception runs through investigation, repair, verification and closure. Re-scoring only asks the model again.
2. Which source in the book's list does the book explicitly decline to treat as proof of a final operative instrument?
- A. The Petroleum Industry Act 2021 [2]
- B. The National Data Repository Regulation on the NUPRC draft register [13]
- C. The Nigerian Oil and Gas Industry Content Development Act [4]
- D. The Nigerian Upstream Petroleum Safety Regulations 2024 [3]
B. The National Data Repository Regulation on the NUPRC draft register [13] — Source [13] is on NUPRC's draft regulations page and the book says it is not proof of a final operative instrument. The safety regulations [3] are on the gazetted register; confusing the two is the trap Chapter 19 warns against.
3. How does Chapter 19 say Nigerian capability should be measured?
- A. By the share of local staff on the payroll
- B. By training hours per employee
- C. By attendance certificates held
- D. By demonstrated work: task, assessment method, supervision conditions and evidence
D. By demonstrated work: task, assessment method, supervision conditions and evidence — The book says to measure capability as demonstrated work, not simply local headcount or an attendance certificate. Payroll share and training hours are not demonstrated work either.
Today’s Assignment
Write Section 8 of your First-Year AI Operating Programme: your company's protection boundary, one obligation held as a controlled record, and how Nigerian content and local knowledge will be made measurable.
- State the boundary between prediction and protection in your business, and what happens if a model is wrong or unavailable.
- Name one record in your company where a status could change without evidence today: a hazard, an environmental exception or a community commitment. Say how you will fix it.
- Take one regulatory obligation that applies to your company and write it as a controlled record: title, issuer, status, effective date where confirmed, version, applicability, owner, the task and the evidence. Use only the instrument itself, and keep any draft labelled draft.
- Describe how Nigerian content and capability are measured in your business today, and how you will make them measurable as demonstrated work.
- Say what knowledge would leave if your main technology supplier left, and the rights you will require.
- Be honest about what your company currently gets wrong. If your company has no AI in use yet, work from the obligations and records in front of you. Write 400 to 900 words.
YOUR 90-DAY PLAN
Section 8 — Our protection boundary, our obligations as controlled records, and measurable Nigerian capability.
How this is marked — the rubric
| States a clear protection boundary with the response if the model is wrong or unavailable | 25 |
| Writes one obligation as a controlled record with every required field and no invented requirement | 25 |
| Defines Nigerian capability as demonstrated work with task, assessment, supervision and evidence | 20 |
| Identifies knowledge-retention rights and one record status that can change without evidence | 15 |
| Gives an honest account of what the company currently gets wrong | 15 |
| Total · pass mark 70 | 100 |
Submitting session 8
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: 8, 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 2 — Services, Safety and Capability 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 2 →Closed book, no notes, and Coach Ritchie switches himself off while you sit it.Capabilities Covered
Protection boundaryHonest environmental recordsCommunity commitmentsControlled obligationsMeasurable Nigerian capability
Part Four–Five · Control and ValueSESSION 9
Generative AI, Cybersecurity and the Investment Case
Why This Matters
The book judges generative AI by task, not fluency: a verbose answer that sounds professional but misses the omission is not successful. Today puts three disciplines around it. First, an assistant or agent must be bounded: what it can read, draft, approve and execute. Second, every system needs a lifecycle with an owner, monitoring, security and a planned end. Third, the investment case must name the mechanism by which value arrives and the design that will prove it.
25-Minute Module
0:00 – 6:00
Retrieval built around authority
- Retrieval-augmented generation combines a language model with retrieved information: the system looks up documents, then writes its answer from what it found. The foundational method is a useful reference [8], but an enterprise assistant also needs access controls, version control, source selection, retrieval evaluation and a response policy. Retrieval does not eliminate hallucination or guarantee authority.
- Apply authorisation before returning content, so a summary never bypasses a user's inability to open a document. Track whether a passage is approved, superseded or informational.
- Evaluate the complete answer chain, retrieval separately from generation: was the necessary source retrieved, were its conditions preserved, did the cited passage support the claim, and did the system refuse when the source was absent or conflicting?
- THE DOCTRINE
Measure outcomes by task rather than by fluency. A verbose response that sounds professional but fails to detect the omission is not successful.
- Keep a versioned evaluation set that grows from actual failures.
When a vendor shows you a beautiful answer, ask to see the question it refused.
6:00 – 11:00
Agents with bounded tools
- An agent combines reasoning with tool use. Define the allowed tools, schemas, identities, transaction limits and approvals. Read access, draft creation and release permissions are distinct.
Permission Chapter 20's tender agent Read access A permission of its own, granted separately from drafting or release Draft creation May prepare a draft compliance matrix Release Must not submit a bid, change a price or claim a qualification without authorisation - Treat retrieved content as untrusted input. A document containing an instruction to send files externally must not become an instruction to the agent. Restrict destinations and log every proposed tool call and approval.
- Send arithmetic to deterministic code; reject invented identifiers and incompatible units. Re-evaluate after any change in model, prompt, index, tools or policy, and keep an operable rollback version.
- THE DOCTRINE
Ask what the agent can read, draft, approve and execute. Require a demonstration that a malicious document cannot expand those permissions.
Ask for the agent's permissions on a single page: what it can read, draft, approve and execute.
11:00 – 16:00
Lifecycle, drift and security
- Chapter 21: the deployed system includes data sources, transformations, model, prompts, index, interfaces and users. Version all of them, and assign owners for availability, quality, security and change control.
- The NIST AI RMF provides voluntary governance functions for identifying and managing AI risk; the book applies its ideas to a proposed operating process and does not treat it as a Nigerian statutory obligation [1]. The OT security guide, NIST SP 800-82 Revision 3, addresses systems that interact with physical operations [6]. An initial public draft of Revision 4 exists; a draft must not be treated as a final standard.
- Separate drift from performance failure. Input drift means the data changed; performance deterioration means the system became less useful or accurate. One can occur without the other, so drift prompts investigation, not automatic retraining.
- Personal information in personnel, coaching and medical records needs a lawful, proportionate and secure process under the applicable framework; the book identifies the Nigeria Data Protection Act 2023 [5]. Removing names does not necessarily anonymise.
- THE OPERATOR · THE WORKED CASE
- Operator. A fictional operator's document assistant serves its engineers and contractors. Authorisation applies before content is returned, so a contractor never sees an operator document it cannot open.
THE SERVICE COMPANY · THE PARALLEL- Service company. The fictional contractor retires its quote-review assistant when a better process replaces it, revoking credentials so no forgotten scheduled job keeps recommending.
- Appendix D's model incident review: preserve versions and logs, classify the issue (data, domain, modelling, retrieval, tools, user interpretation or execution), suspend if necessary and re-evaluate before restoring. A prompt change alone may not fix an access-control failure.
Continuing to run is a decision that needs evidence just as deployment does. Put it on the agenda as one.
16:00 – 22:00
The investment case and the counterfactual
- Chapter 22: an investment case explains how information changes an action and how that action changes a business result. A quote reviewer might reduce omissions; a reliability warning might change inspection timing.
Benefit category What the book warns Cash savings Staff hours released do not become payroll savings unless the cost actually changes Capacity released Recognise it accurately; do not relabel it as cash Additional contribution Additional production creates value only if it is feasible, accepted and economically realisable Risk reduction Needs a defensible consequence and probability model before a monetary figure - Count the complete cost, tested against plausible adoption levels. The simplified annual value is incremental contribution plus realised savings minus recurring programme cost, and it is no substitute for a full investment model where capital, taxes or contractual exposure are material.
- Choose the evaluation design before rollout. Appendix G frames the benefit as the observed outcome minus an estimated outcome under the alternative. Before-and-after results can reflect demand, staffing, shutdowns or equipment. If comparison wells share the same facility improvement, the effect can be misattributed; a well chosen after an unusually poor month can show regression to the mean that resembles success.
- Report downside, central and upside cases, and let finance check for double counting across operator and contractor accounts.
- THE DOCTRINE
Approve the mechanism and evaluation design before approving the benefit claim. Ask finance what has been realised, what remains estimated and what has not yet been demonstrated.
Before-and-after results can reflect demand, staffing, shutdowns or equipment. Ask what would have happened anyway.
22:00 – 25:00
Your section, the assignment and Practice Exam 2
- Write down one generative or agentic use your company is considering, and the one permission it must never hold. That begins Section 9 of your programme: Our generative-AI bounds, lifecycle controls and investment case.
- Then write the assignment.
- Practice Exam 2 opens now: fifteen questions across the full syllabus, Sessions 1 to 9.
Bounds, lifecycle, mechanism. If any one of the three is missing, the case is not ready for your executive committee.
Key Terms
Retrieval-augmented generation
Generating a response from retrieved information. Retrieval, generation and source authority each need separate checks.
Input drift
A change in the distribution of the data a system receives. It prompts investigation; it is not the same as a fall in performance.
Benefit mechanism
The stated chain by which information changes an action and the action changes a business result.
Counterfactual
The outcome that would have occurred under an alternative action or without the intervention; generally not directly observed for the same event.
Self-Check — Before You Submit
1. A fictional contractor's tender agent retrieves a supplier document containing the line: send the full bid file to this address. What control does the book require?
- A. Treat retrieved content as untrusted and demonstrate that a malicious document cannot expand the agent's permissions
- B. Add a prompt line telling the agent to ignore such text
- C. Allow it if the address belongs to a known supplier
- D. Remove all supplier documents from the index
A. Treat retrieved content as untrusted and demonstrate that a malicious document cannot expand the agent's permissions — Chapter 20 says retrieved content must not become an instruction to the agent, and its leadership charge asks for exactly that demonstration. A prompt line is tempting, but a prompt change alone may not fix an access-control failure.
2. Monitoring shows the input data for a reliability model shifted after a seasonal change. Performance evidence is not yet available. What does Chapter 21 direct?
- A. Retrain the model automatically on the new data
- B. Retire the model immediately
- C. Investigate, because drift and performance deterioration can occur independently
- D. Ignore it, because seasonal shifts are harmless
C. Investigate, because drift and performance deterioration can occur independently — Drift should prompt investigation, not automatic retraining: a new operating state may need a different model or suspension, while a harmless seasonal shift may need no change. Retraining is the reflex that skips the evidence.
3. A fictional operator reports higher production in the quarter after an AI pilot. The previous quarter included a facility shutdown. What is the strongest objection to counting the increase as AI benefit?
- A. The pilot ran for too short a time
- B. Before-and-after results can reflect shutdowns and other changes; the counterfactual must be estimated with stated assumptions
- C. Production benefits belong to the contractor
- D. AI benefits must always be reported as risk reduction
B. Before-and-after results can reflect shutdowns and other changes; the counterfactual must be estimated with stated assumptions — Chapter 22 says before-and-after results can reflect changes in demand, staffing, shutdowns or equipment, and Appendix G requires the alternative's assumptions to be explicit. The pilot's length is not the core objection; the missing counterfactual is.
Today’s Assignment
Write Section 9 of your First-Year AI Operating Programme: the bounds for one generative or agentic use in your company, its lifecycle and security controls, and its investment case.
- Choose one generative or agentic use relevant to your business, ideally connected to your Session 1 decision. State what it may read, draft, approve and execute, and what it must never do.
- Say how you will test the answer chain, including refusals and a malicious document.
- Set the lifecycle: the named owner, what is versioned, how drift and performance are monitored separately, the access and personal-data controls, and the conditions for suspension and retirement.
- Write the investment case: the benefit mechanism, the benefit category, the complete cost, and the evaluation design you will choose before rollout, with its counterfactual assumptions.
- Give downside, central and upside cases, and state what has not yet been demonstrated.
- Be honest about what your company currently gets wrong. If your company has no AI in use yet, work from the use being proposed to you. Write 400 to 900 words.
YOUR 90-DAY PLAN
Section 9 — Our generative-AI bounds, lifecycle controls and investment case.
How this is marked — the rubric
| Bounds the use precisely by read, draft, approve and execute, with a malicious-document test | 25 |
| Names an owner and sets monitoring, security, personal-data and retirement controls | 25 |
| States the benefit mechanism and category accurately with a complete cost | 20 |
| Chooses an evaluation design before rollout with explicit counterfactual assumptions and three cases | 20 |
| Gives an honest account of gaps without claiming unevidenced savings | 10 |
| Total · pass mark 70 | 100 |
Submitting session 9
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: 9, 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 2 — Full Syllabus 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 2 →Closed book, no notes, and Coach Ritchie switches himself off while you sit it.Capabilities Covered
Bounded agentsAnswer-chain evaluationModel lifecycle controlData securityInvestment case disciplineCounterfactual thinking
CapstoneSESSION 10
The First-Year Operating Programme
Why This Matters
Nine sessions ago you named your company, its side of the operator–contractor relationship and one decision whose loop you would map. Every section since has been written around that decision. Today you assemble the nine sections into one document, organise it as a first year of work, give each part a role that can carry it, and sign it: your First-Year AI Operating Programme. Chapter 24 calls its sequence a proposed programme, not a universal completion promise. Then you sit the Final Exam, and we close where the book closes: with the leader.
25-Minute Module
0:00 – 5:00
Roles people can use
Role (Chapter 23) What it holds Executive sponsor Resolves cross-functional barriers Domain lead Defines the decision and the technical boundary Data steward Maintains definitions and quality Model and application owners Maintain the system Security, compliance and finance Provide relevant assurance Action owner Remains responsible for use in the operating process - Small organisations can combine roles, but no one person should design, validate and approve a high-consequence application without independent review.
- Teach judgment rather than prompting alone, using examples that include failure, such as an obsolete procedure or an out-of-domain operating state. Assess whether the user knows when to stop and escalate.
- Place the output where the decision already occurs: a quote exception belongs in the approval workflow. Keep reason codes for overrides, and do not make the model team responsible for all operating results.
- THE DOCTRINE
Adoption is demonstrated in decisions and outcomes, rather than attendance at a launch event.
Encourage people to report failures promptly. Treat honest uncertainty differently from deliberate misrepresentation, and say so out loud.
5:00 – 11:00
The first year, phase by phase
- Chapter 24 starts with a small decision portfolio, a measured baseline and one low-consequence administrative use and one technical advisory use. Service companies connect opportunity qualification to quotation assumptions and mobilisation readiness; operators connect a technical opportunity to the engineer and operating process that can act.
Phase What happens What the leader decides Starting point (first month) Map decisions and owners; success and suspension criteria; data contracts and source register Which two uses start, and on what baseline Build and evaluate in bounded phases Baselines and candidates on withheld data; fix structural data problems first; shadow operation Whether the evidence supports controlled use Introduce controlled use Integrate into the workflow after authorised review; log outcomes; test outages and rollback At midyear: improve, expand, pause or stop each application Scale through reusable controls (second half) Reuse data contracts, evaluation methods, approval records and integration patterns where they fit; revalidate transfer to new wells, equipment or documents At year end: realised value, unresolved risk, lifecycle cost and capability developed - THE DOCTRINE
A calendar month with no relevant failures cannot establish the reliability of a rare-event warning system. Record what was learned and what remains untested.
- A pause can be the right result. Do not expand merely to show more use cases, and let year-end evidence shape next year's portfolio.
Set the pace by evidence and consequence. An administrative use can move quickly; a decision connected to well integrity takes longer. That reflects consequence, not resistance to innovation.
11:00 – 16:00
Rehearse it jointly: the book's own exercise
- Appendix H is hypothetical: a mature well with declining accepted oil rate and more interruptions, and a contractor whose diagnostic capacity is shared. The first joint decision is not which AI model to buy. It is whether the evidence supports a diagnosis and a feasible next action.
- Seven steps: establish the evidence (two reports define interruptions differently, resolved before modelling); define competing explanations; compare feasible actions; construct the quotation; execute and capture evidence; assess outcomes and economics; update the operating memory.
- THE OPERATOR · THE WORKED CASE
- Operator. The technical owner defines the decision and engineering boundary, then reconciles technical effectiveness, availability and economic contribution, estimating the counterfactual with explicit limitations.
THE SERVICE COMPANY · THE PARALLEL- Service company. The operations owner defines resource readiness; the contractor reconciles costs, authorised revenue, invoice acceptance and cash timing. Its avoided standby and the operator's accepted production stay distinct, never one unexplained savings figure.
- Workshop outputs: decision record, data contract, competing hypotheses, evaluation plan, scope-to-cost map, mobilisation checklist and outcome record.
- THE DOCTRINE
The exercise is complete only when learning affects the next job or decision.
If you can, run Appendix H with your counterpart across the contract.
16:00 – 21:00
Assemble the nine sections and answer Appendix E
- Foundations. Section 1 — Our company, our side of the operator–contractor relationship, and the one decision whose loop we will map. Section 2 — Our binding constraint, decision portfolio and field data contract. Section 3 — The credible alternative, the model task, the abstention policy and the evaluation design.
- The oilfield. Section 4 — Where subsurface, reservoir or drilling intelligence changes an action in our business, and what information is worth buying. Section 5 — Our field application: the system it optimises, the alert policy, the integrity evidence and degraded operation.
- Services, trust and value. Section 6 — Our tender, quote and mobilisation decisions and the controls they need. Section 7 — From plan to collected cash: our execution evidence, job economics and reporting discipline. Section 8 — Our protection boundary, our obligations as controlled records, and measurable Nigerian capability. Section 9 — Our generative-AI bounds, lifecycle controls and investment case.
- Keep that order, one heading per section, laid across Chapter 24's phases, with each Chapter 23 role named by job title. On top goes a one-page summary: decision, two first uses, owners, baseline, success and suspension criteria, review dates.
- Answer the Appendix E questions for your role: board and managing director, technical director, service company operations lead, finance and commercial leadership, or assurance and information security. The board, for example, asks which benefit has been realised and which remains estimated. A demonstration's fluency does not answer it.
Where a later section changed your mind about an earlier one, correct it and say so. That is rigour, not inconsistency.
21:00 – 25:00
Sign, sit the Final Exam, close the loop
- Sign and date it; the committee may then hold you to its owners and dates.
- The Final Exam opens now: fifteen questions, twelve from the paper and three rebuilt from your own earlier mistakes, closed book, pass mark 70.
- The conclusion is a chain: reservoir knowledge becomes an engineering choice, then an authorised job, then accepted records; value reaches the bank; learning shapes the next decision.
- THE DOCTRINE
The responsibility stays with the leader. Require evidence. Preserve uncertainty where it matters. Give people usable authority. Make the result measurable. Close the loop.
- The book opens with a personal note from me, and I will close the programme with it: AI can improve an operator's decisions. Leadership must turn those decisions into safe operations, reliable production and lasting value.
It can begin with one carefully bounded decision. It becomes an operating doctrine when the same discipline is applied across wells, equipment, jobs and commercial commitments.
Key Terms
First-Year AI Operating Programme
Sections 1 to 9 assembled across Chapter 24's phases, with Chapter 23 roles named, a summary and a signature.
Action owner
The person responsible for using a system's output in the operating process. The model team does not inherit that responsibility.
Shadow operation
Running a system alongside the existing process without allowing its outputs to change the operational action.
Midyear review
The point at which each application is improved, expanded, paused or stopped on evidence.
Self-Check — Before You Submit
1. At a fictional contractor's midyear review, a warning system has run in shadow for one month with no relevant failures. The sponsor wants to expand it to five more equipment classes. What does Chapter 24 support?
- A. Expand now, because a clean month shows reliability
- B. Expand to two classes as a compromise
- C. Stop it, because it has caught nothing
- D. Do not expand yet: continue shadow operation and record what remains untested
D. Do not expand yet: continue shadow operation and record what remains untested — A calendar month with no relevant failures cannot establish the reliability of a rare-event warning system, and the book warns against expanding merely to show more use cases. Stopping is also premature: a pause is a legitimate result.
2. In Appendix H, how should the contractor's avoided standby and the operator's additional accepted production be reported?
- A. As distinct benefits, each reconciled by the party that holds it
- B. As one combined savings figure
- C. Only on the operator's side, since it owns the well
- D. Only after converting both to staff hours
A. As distinct benefits, each reconciled by the party that holds it — Step 6 says both may matter but must not be merged into one unexplained savings figure. A combined figure looks tidy for a board, which is the trap.
3. Users override a new quote-exception tool in a high share of cases. What does Chapter 23 say this signals?
- A. The users need prompt-writing training
- B. The tool should act without human approval
- C. A reason to investigate, using reason codes and narrative evidence
- D. The tool should be withdrawn at once
C. A reason to investigate, using reason codes and narrative evidence — An override may reveal missing context, model error, a misunderstanding or an incentive conflict. Chapter 23 says a high override rate is a signal to investigate, not a reason to remove human authority automatically.
Today’s Assignment
Assemble Sections 1 to 9 into one signed First-Year AI Operating Programme for your company's executive committee, organised by Chapter 24's phases and Chapter 23's roles, with your answers to the Appendix E questions for your role.
- Assemble the nine sections in order, noting in one line where a later session changed an earlier one.
- Lay the programme across Chapter 24's four phases, with the midyear decision and year-end report.
- Name each Chapter 23 role by job title in your company, and how independent review is supplied.
- Write the one-page executive summary described in this session.
- Answer the Appendix E questions for your role with evidence, marking honestly any you cannot yet answer.
- Sign and date it, naming the commitment you are least confident your company will keep. The programme will exceed 900 words; the summary may not. With no AI in use yet, it still stands: it starts from the decision.
YOUR 90-DAY PLAN
The assembled, signed First-Year AI Operating Programme.
How this is marked — the rubric
| All nine sections present, in order and consistent, with corrections noted | 25 |
| Organised across Chapter 24's phases with midyear and year-end decisions and suspension criteria | 20 |
| Chapter 23 roles named by job title with proportionate independent review | 20 |
| Appendix E questions for the student's role answered with evidence or honest gaps | 20 |
| One-page summary, signed, with the least-certain commitment named and explained | 15 |
| Total · pass mark 70 | 100 |
Submitting session 10
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: 10, 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.FINAL EXAM
Your Final Exam is sat in this session. 15 questions, of which 3 are rebuilt from questions you got wrong on the practice papers. Sit it once your capstone is submitted.
Sit the Final Exam →Closed book, no notes, and Coach Ritchie switches himself off while you sit it.Capabilities Covered
Programme assemblyRoles and accountabilityPhased deliveryJoint operator–contractor workingExecutive reviewClosing the loop