Quick answer: A Unit 4004 project management plan for an AI SOP generator is the Phase 1 document that justifies your project, sets measurable success criteria, compares options, and schedules the work with a Gantt chart and critical path. Write the eight sections in order, and make every target something you can test later.
You have picked the AI-assisted SOP generator, or you are about to, and the brief now asks for a plan before you have written a line of code. That feels backwards. It is not. The Unit 4004 project management plan is where most of the marks for planning, and a surprising share of the marks for the final report, are quietly won or lost. Markers read your plan again in Phase 3 to check whether you did what you said you would do.
This guide explains how to build that plan for the 2026-27 AI theme, section by section, using the SOP generator as the worked example. You can read the Unit 4004 AI theme brief alongside it.
What the Unit 4004 project management plan actually has to do
Unit 4004 Managing a Professional Engineering Project sits in the Pearson BTEC Higher National in Engineering at Level 4. For 2026-27 the Pearson-set theme is artificial intelligence, and the assignment asks you to investigate how far AI can improve an engineering or manufacturing documentation process without unacceptable cost, error, extra review work or disruption.
The suggested project is a generator for standard operating procedures. The user supplies a company template and process information, the application drafts the SOP, points out what is unclear or missing, and the user reviews, edits and exports it. Two rules shape everything: a human must review the output, and the tool must not invent missing information.
Your plan has one job. It must show that you understand the problem, have a testable question, and can deliver a working application and an honest evaluation inside the time you have. It is not a place to describe the technology at length.
The eight sections and where the marks sit
| Plan section | What to write | Typical weak point |
|---|---|---|
| 1. Introduction and justification | The documentation problem and why AI is worth testing | Generic praise of AI |
| 2. Stakeholders and baseline | Who writes, reviews and uses SOPs; current time and error figures | No baseline measured |
| 3. Question, aim, objectives | One question, one aim, SMART objectives | Vague objectives |
| 4. Scope and success criteria | In and out of scope; measurable acceptance criteria | Criteria that cannot be tested |
| 5. Feasibility study | Alternatives, costs, resources, timescales, risks, assumptions | Only one option considered |
| 6. Research methodology | Primary and secondary methods, sampling, instruments, analysis | Methods not linked to objectives |
| 7. Planning and management | Gantt chart, critical path, milestones, logbook | Chart with no dependencies |
| 8. Legislation, ethics, risk | Relevant law, ethical principles, risk register | Risks copied from a template |
Notice the pattern in the right-hand column. Almost every weakness is a failure to be specific. That is good news, because specificity is a habit you can build.
Step-by-step: building your project management plan
- Measure the baseline first. Time yourself, or a colleague, writing an existing SOP by hand. Count reviewer corrections. Without this, “improvement” means nothing.
- Write one research question. The brief gives you the territory: to what extent can AI consistently and reliably improve documentation processes? Narrow it to your SOP case.
- Turn the question into four to six SMART objectives. Each one gets a measure and a week number.
- Set acceptance criteria with numbers. For example, a target for time saved, a target for correct steps, a target for flagged gaps, and a rule that any invented content fails a test run.
- Compare at least three options. Manual drafting, a bought tool, and your own build is a sensible set. Score them on cost, time, risk and learning value.
- Choose methods that produce evidence. Timed trials, a reviewer scoring sheet and a short user questionnaire cover time, accuracy and usability.
- Build the schedule from dependencies. List tasks, durations and what each depends on, then find the longest chain. That is your critical path.
- Finish with law, ethics and risk. Keep this specific to documents that control how people work safely.
If you want a second pair of eyes before the draft plan is reviewed, our HNC engineering assignment help team can look at structure and wording with you.
Scope and success criteria for an AI SOP generator
This is the section students rush, and it is the one the rest of the unit depends on. A success criterion must be measurable, agreed with a stakeholder and testable with the time and tools you have.
Think about what could actually go wrong with the product. A language model can write a confident, well-formatted step that has no basis in your input. In a procedure that tells someone how to isolate a machine, that is a safety problem. So your criteria should treat invented content as a failure in its own right, not as a small dent in an accuracy percentage.
Good criteria for this project usually cover five things: time saved including review time, accuracy of steps against the source notes, how many deliberately removed details the tool flags, whether repeat runs on the same input give a consistent structure, and a usability rating from the people who will use it. Set the numbers yourself, justify them, and be ready to report honestly if you miss them. Missing a target is not a fail. Hiding it is.
Feasibility, methodology and the vibe coding question
The brief encourages AI-assisted development, sometimes called vibe coding, with tools such as GitHub Copilot and Microsoft Azure for Students for hosting. These come with limits on usage and credit, and the brief expects you to manage tools, costs and testing against requirements. So the feasibility section should say what you will use, what it might cost, what happens when a limit is reached, and what you will do instead.
Be careful about one thing. Using AI to write code does not mean you can skip understanding it. Keep version control, record the prompts that mattered, and be able to explain how your application handles a missing field. In the professional discussion you will be asked.
For methodology, link each method to an objective. A timed trial answers the time objective. A scoring sheet answers accuracy. A questionnaire answers usability. If a method does not serve an objective, drop it. With a small sample, say plainly that your findings are indicative, not statistically conclusive. Markers respect that sentence more than inflated confidence.
You can see these sections written out in a worked Unit 4004 AI SOP generator sample (for reference only, not for submission as your own work).
Gantt chart and critical path analysis
A Gantt chart that is only a row of coloured bars is decoration. What turns it into a management tool is dependency. Which tasks cannot start until another finishes? Which can run in parallel? The chain of dependent tasks with the longest total duration is the critical path, and any delay on it delays the whole project.
For an SOP generator, the chain usually runs from baseline measurement, to requirements, to build, to first test round, to the improvement cycle and retest, to evaluation and report writing. Ethics checks and data preparation can often run beside the build, so they have slack. Mark the milestones the unit gives you: the proposal and feasibility review, the logbook review and the interim report. Then keep a dated logbook from day one. A logbook written in one evening at the end is easy to spot and earns little.
Build in time for a second improvement cycle. The brief asks for iterative testing and retesting, so a plan with a single test round does not match what you are being asked to do.
Legislation, ethics and risk
Keep this section tied to the project. Standard operating procedures are safety documents, so the Health and Safety at Work etc. Act 1974 is relevant: an AI draft must never reach the shop floor before a competent person has approved it. If your test inputs include any personal data, the Data Protection Act 2018 and UK GDPR apply, and the safest route is to anonymise everything and say so. The Engineering Council and Royal Academy of Engineering Statement of Ethical Principles gives you a professional frame: honesty, accuracy and rigour, and responsibility to society.
For the risk register, avoid generic entries. Use risks that belong to this build: the model invents a step, output varies between runs, a usage or credit limit stops testing, confidential material is sent to an online model, the reviewer is unavailable. For each, give likelihood, impact and a mitigation you can actually carry out. Human review is the standing control, and your plan should say who does it and when.
Mark-maximising checklist
- A measured baseline, with the method described.
- One research question and objectives that each have a number and a date.
- Acceptance criteria that include a rule about invented content.
- At least three options compared, with a reasoned choice.
- Methods mapped to objectives, with a sampling limitation stated.
- A Gantt chart showing dependencies, a named critical path and milestones.
- A risk register specific to AI-assisted SOP drafting.
- A dated logbook that records decisions, failures and plan changes.
- Harvard references that you have checked exist.
Common mistakes
The first is writing about AI in general instead of about your project. The second is setting targets you never test. The third is a plan that is never revisited: when your first test round goes badly, record the change and the reason, because reviewing plan changes is part of Phase 2. The fourth is treating the report as a separate task, when the plan, logbook and results should clearly connect. The fifth is leaving referencing until the end and finding that a source cannot be traced.
Finally, remember who is responsible for the work. Model answers and guidance help you understand structure, but your plan must describe your own project, your own data and your own tests, and you must follow your centre’s rules on AI use. Read what students say in our reviews if you want to see how others have used this kind of support.
FAQs
What is the Unit 4004 project management plan?
It is the Phase 1 document in which you justify your project, set your question and objectives, define success criteria, assess feasibility, choose research methods, schedule the work and manage legislation, ethics and risk. It guides everything you do in the later phases and is checked again when you write your report.
Do I have to choose the AI SOP generator project?
No. The SOP generator is the suggested project, and it fits the brief well because it involves documentation, human review and measurable benefits. You may propose another project if it meets the brief and is approved by your centre, but check this with your tutor first.
How detailed should the Gantt chart be?
Detailed enough to show tasks, durations, dependencies and milestones across the project period. If a reader cannot work out your critical path from the chart, it needs more structure. Fifteen to twenty well-defined tasks is usually more useful than a handful of vague bars.
Can I use GitHub Copilot to build the application?
The brief expects AI-assisted development using student tools. You should still understand and be able to explain your code, keep a record of how AI helped, and test the result against your requirements. Declare your AI use in line with your centre’s guidance.
What counts as a good success criterion?
One that is measurable, linked to a stakeholder need and testable within your timescale. For example, a percentage reduction in drafting time including review, or a minimum share of deliberately removed details that the tool flags. Vague aims such as “improve efficiency” will not earn marks.
What if my AI tool does not meet my targets?
That is still a valid result. Report the evidence, analyse why the targets were missed, show what you changed, and give recommendations. An honest evaluation of a partial success usually scores better than an overstated claim of success.
If your draft plan is due and the sections are not fitting together, you can send your brief through our assignment submission page for a model answer and structured support, or explore our HNC assignment help service first.
About the expert
This guide is presented under the review of Dr. John Evans, who holds a PhD in Classical Mechanics from ETH Zurich and specialises in Classical Dynamics, Celestial Mechanics and Fluid Mechanics.
