Level 1 · Reaction
80 %of learners rate the course useful for their daily work.
- End of course
Majlis Academy · Bootcamp · Open now
ADLC bootcamp
AI in the engineering delivery lifecycle
Most engineering teams use AI only while they write code. They do not know what each task costs, and they cannot prove that AI-assisted work meets the requirement. In four days, your team learns to use AI from intake to release, at a known cost, with evidence and human approval at each gate.
- Format
- Four-day classroom bootcamp, at your site
- Time
- 20 learner hours, 66 % hands-on practice
- For
- Engineers, QA, tech leads and platform teams
- Group
- 12 to 24 learners, in teams of three or four
Aligned with
The questions we answer
Three questions your team returns to all week.
#Q1
Where in our delivery lifecycle does AI add value, and where does it only add cost?
Answered onDay 1: baseline, token economics, specification
#Q2
What evidence proves that AI-assisted work meets the requirement?
Answered onDays 2 and 3: context, traceability, verification agent
#Q3
Which decisions must a person keep, and why?
Answered onDays 3 and 4: approval gates, readiness, governance
01 · The ADLC reference model
Three layers that stay the same when tools change.
The lifecycle your engineers already know, how AI takes part in each stage, and what makes it safe in a regulated institution. Tools and models change every quarter. The model does not.
Delivery
- Requirement intake
- Analysis and decomposition
- Planning
- Implementation
- Testing and verification
- Release
- Change and improvement
AI execution
- Prompts as specifications
- Context and repository grounding
- Model routing
- Tool use
- Specialist agents
- Review loops
Governance
- Token and cost control
- Security and compliance
- Human approval gates
- Quality evidence
- Production readiness
- Audit and accountability
the lifecycle you know, with AI inside it and a gate on each stage
02 · Four days, in lifecycle order
Four days. Five learner hours each.
Each day follows the lifecycle: from the baseline on day 1 to governance and adoption on day 4. No course at the Academy runs longer than four days. Two thirds of every day is practice on your own repositories or our synthetic Sandbox Bank.
- 1
Baseline, token economics and specification
Map your team's AI use, record the cost of real tasks, and turn prompts into specifications.
→ 3.25 h practice · 1.75 h input
- 2
Context, planning and agentic implementation
Ground the repository. Decompose a requirement with approval points. Run a plan-then-execute agent flow.
→ 3.5 h practice · 1.5 h input
- 3
Verification and production readiness
Trace acceptance criteria to tests. Build a verification agent. Decide go or no-go with evidence.
risk, security and architecture join
→ 3.5 h practice · 1.5 h input
- 4
Governance, loops and adoption
Intake checklist, review agent, a controlled feedback loop and the adoption plan.
risk, security and architecture join
→ 3 h practice · 2 h input
03 · Twelve outcomes
Twelve outcomes. Each one practised and checked.
After the course, the learner can do each of these. Every outcome has a graded practice block, and every practice block produces an asset your team keeps.
| # | After the course, the learner can | Level and practice | Time |
|---|---|---|---|
| Day 1 · Baseline, token economics and specification | |||
| 01 | Map AI use and record a delivery baselineMap your team's AI use onto the ADLC reference model, with the time, quality and cost of three real tasks. | AnalyseArchiMate modelling | 0.75 h |
| 02 | Design a model routing rule setDesign a model routing rule set for five task types and show the change in cost per task against the baseline. | CreateHands-on case | 1.25 h |
| 03 | Rewrite a prompt as a specificationRewrite a prompt from your team as a specification-grade prompt with acceptance criteria, an output contract and a verification check. | ApplyGallery walk | 1.25 h |
| Day 2 · Context, planning and agentic implementation | |||
| 04 | Configure repository contextConfigure the project context of one repository with a rules file and a context-loading pattern that keeps token use within a stated limit. | ApplyHands-on case | 1 h |
| 05 | Decompose a requirement with approval pointsBreak down one requirement into delivery, implementation and verification tasks, and mark the approval point for each task. | AnalyseHands-on case | 1 h |
| 06 | Run a plan-then-execute agent flowRun a plan-then-execute agent flow on a code change with defined stop conditions, escalation points and guardrails. | ApplyBranching case | 1.5 h |
| Day 3 · Verification and production readiness | |||
| 07 | Trace acceptance criteria to testsTrace each acceptance criterion of a requirement to its verification checks, and find the gaps and the noisy tests. | AnalyseHands-on case | 1 h |
| 08 | Build a verification agentBuild a verification agent that checks a pull request against its acceptance criteria and test evidence, and escalates defined cases to a person. | CreateHands-on case | 1.5 h |
| 09 | Assess production readinessAssess an AI-assisted change against a production readiness definition and justify a go or no-go decision with evidence. | EvaluateReview-panel role-play | 1 h |
| Day 4 · Governance, loops and adoption | |||
| 10 | Design an intake checklist and review agentDesign an AI intake checklist and a review agent that classifies each control as pass, fail or escalate. | CreateHands-on case | 1.25 h |
| 11 | Design a controlled feedback loopDesign one feedback loop for a real process with an evaluation signal, a stop condition and a cost boundary. | CreateHands-on case | 1 h |
| 12 | Justify an adoption and ownership planJustify an adoption plan that names the owner, the next step and the sponsor of each artefact that the course produced. | EvaluateWardley mapping | 0.75 h |
04 · What your team keeps
Working assets, not slides.
Every outcome produces an asset that your team owns after the course. We build them on your repositories and planning system where your policies allow, or on our synthetic Sandbox Bank.
- 01ADLC reference model view
- 02Baseline task set
- 03Cost-per-task comparison
- 04Routing rule set
- 05Specification-grade prompts
- 06Repository context pattern
- 07Decomposition example
- 08Requirement-to-planning flow
- 09Agent flow and run log
- 10Traceability map
- 11Verification agent
- 12Production readiness definition
- 13Intake checklist
- 14Intake review agent
- 15Feedback loop design
- 16Adoption and ownership plan
05 · How we measure success
Four levels. Four targets.
Kirkpatrick levels 1 to 4. The first two are measured at the end of the course, the last two on your live work afterwards.
Level 2 · Learning
85 %of learners pass the graded work of all twelve outcomes.
- End of course
Level 3 · Behaviour
2+teams use the routing rules and the verification agent on live work.
- After 60 days
Level 4 · Results
Loweraverage cost per task against the day 1 baseline, with review quality held.
- After 90 days
Programme targets, not reported results.
06 · Practical details
Practical details
- Who should attend
- Software engineers, QA engineers, tech leads and platform engineers. Risk, security and architecture colleagues join on days 3 and 4.
- Before the course
- Daily use of Git, a CI pipeline and a planning system. Some use of an AI coding assistant.
- Delivery
- Classroom, at your site. Four days of 5 learner hours, with micro-lessons of 6 to 12 minutes between practice blocks.
- Language
- English, with an English–Arabic glossary of ADLC terms.
- Tools and data
- Only the AI tools and models that your organisation approves. Exercises use synthetic data unless your data owner approves otherwise.
- Your sector
- We adapt the cases to banking (BIAN), telco (TM Forum ODA, eTOM, SID) or government.
- After the course
- A one-page ADLC job aid, two office-hours sessions at 2 and 6 weeks, and a review of your adoption plan at 30 days.
How Majlis Academy builds courses
Fast on top. Solid underneath.
Backward design from measurable outcomes. Hands-on practice for two thirds of the time. Materials in Simplified Technical English. Assets kept as documentation as code, so the course is rebuilt when the tools change and the core stays.
How the Academy worksPart of Majlis Academy
We advise. We train. We build.
- GovKit Core and the three sector tracksGovernment, banking and telco. Short, always-current programmes on a solid core.
#1 · Courses
- Decide fast, with one owner and a control your regulator can testDecision sprints, programme design and design authority.
#2 · Advisory
- AI factories, built with your engineers in your own tenantSoftware, governance, cloud, data, MLOps and training.
#3 · AI Factories
We advise, we train and we build with you.
Agree the gates once. Then deliver with AI at full speed.
Plan a cohort
Bring one team and one real requirement.
Write to us with your sector, the size of your team and the dates you have in mind. We reply with a proposed plan for your environment.