How To Approach The LOG309 ECA: GenAI Prompt, API Design, Video And Slides

Short answer: treat the LOG309 ECA as one connected argument, not four separate tasks. Pick a technology with a clear weakness in Q1, build the Q2 API to fix that exact weakness, reuse your GenAI output in Q3 under two constraints, then let the video and slides repeat the same storyline. Markers reward that thread.

You have a scenario about unpredictable demand, supplier disruption, tariff changes and maritime delays, a GenAI prompt capped at 50 words, a hypothetical API to design, a video of 3 to 6 minutes and a slide deck. Most students lose marks not on knowledge but on disconnection: the prompt says one thing, the API fixes something else and the slides show a third story.

If you would like an expert to check your plan before you start, send your brief for a quote.

What the LOG309 ECA asks you to do

The July 2026 ECA for Industry 4.0 Logistics Applications at the Singapore University of Social Sciences (SUSS) has three sections. The marks tell you where to spend your time.

Check your own brief for the exact wording; this table is a paraphrase to help planning.

Step 1: choose a technology with a weakness you can build on

Do not choose the most fashionable technology. Choose one where the weakness leads naturally into an API. AI-driven demand sensing, a digital supply chain control tower or predictive ETA models all work, because each has a visible blind spot. Demand sensing, for instance, sees orders but not vessel schedules or tariff changes, so the weakness points straight to a supply-side data feed.

A technology with no obvious weakness leaves you inventing one, and that reads as padding. If the scenario mentions three risks and your technology covers only one, say so. That honesty is the weakness.

Step 2: write the GenAI prompt as an instruction set

You have 50 words, so every word must carry an instruction. A strong prompt gives a role, the scenario in compressed form, the task, the structure you want and the output limit. A weak one says “tell me about Industry 4.0 for logistics” and returns an essay you cannot use.

Count the words in your prompt and write the count beside it. Submit the full output, even if it contains an error. Spotting and correcting that error is evidence of critical use, which is what the attribution requirement is testing.

For a worked example with a labelled illustrative prompt, see the LOG309 ECA model-answer extract (for reference only, not for submission). The platform’s guidance on using ChatGPT for assignments in Singapore also helps if you are unsure where your institution draws the line.

Step 3: reflect on the prompt in 50 words

This is only 5 marks, but students waste it with “I could be more specific”. Name three changes and why each helps: add company size and lead times, require stated assumptions, ask for a comparison table before the recommendation. One sentence each is enough.

Step 4: design the API like a one-page document

A mini design document is not code. Markers want to see that you understand how data moves. A table works best because it keeps you inside about 200 words.

1. Name the API and say what question it answers.
2. List inputs and their source systems (ERP, transport management, customs data, the Q1 technology’s output).
3. Describe processing in two or three verbs: validate, estimate, compare.
4. State outputs a planner can act on, such as stockout probability and landed cost per unit.
5. Explain integration: which system calls it and which system receives the result.
6. Add controls: authentication, logging, validation, thresholds for human approval.

Then write the sentence that earns the link marks: this API exists because the Q1 technology cannot do X.

If systems and data design is your weaker side, IT assignment support is available for model answers and feedback.

Step 5: build Q3 around two constraints

You choose two from limited sensor coverage, supplier data quality issues, tariff rule changes and unpredictable maritime delays. Pick the two that your Q1 output and Q2 API already touch, so the story stays tight. Supplier data quality and maritime delays pair well with a digital twin; tariff rule changes and supplier data suit a rules-based landed cost engine.

The phrase to remember is “still gives useful output”. Say how: confidence scores, scenario ranges, fallback averages, ranked actions. A model that stops when data is poor has not adapted. One that widens its range and tells the planner which action holds up across scenarios has.

Step 6: plan the video from the mark scheme

Organisation carries 8 marks, so state the structure in the first 30 seconds. Pace and articulation carry 7, so rehearse out loud and time yourself. Eye contact and posture are 5 each: look at the lens, not the screen, and sit upright with your hands visible.

Aim for about 5 minutes. Most people speed up on camera, so a script that runs 5 minutes in rehearsal may finish at 4 in the recording. Speak to the camera lens, and treat your slides as support rather than a script to read aloud.

Step 7: design slides without a template

The brief rules out internet templates, so build from a blank deck. Use one font family, two colours and one message per slide. A sensible 13 to 15 slide flow runs: scenario, technology, prompt, output, strength and weakness, prompt improvement, API overview, API table, controls, Q3 proposal, constraints, conclusion, attribution and references. Draw your own flow diagram for the API; it is the easiest way to pick up the creativity marks.

Presentation delivery is its own skill, and these presentation tips are a useful read before you record. If you want help building a deck, see PowerPoint presentation writing help.

Mark-maximising checklist

• Prompt is 50 words or fewer, with the word count shown.
• Full GenAI output is included and is 200 words or fewer.
• Tool, version, date and what you adapted are stated.
• Strength and weakness each have a reason, not a label.
• Q2 names inputs, processing, outputs, integration and controls.
• Q2 and Q3 both refer back to the Q1 output.
• Two constraints are named and each has a specific mitigation.
• Video is between 3 and 6 minutes.
• PDF has no more than 15 slides.
• APA 7th references, with GenAI cited.

Common mistakes

Going over 50 words in the prompt is the obvious one, and it costs marks immediately. Less obvious is trimming the GenAI output to make it look better. The brief wants the full output, flaws included.

Another frequent error is describing an API as software code. You are not asked to write endpoints in Python. A table with six rows is better than ten lines of pseudo-code. Students also pick constraints they never use, so Q3 reads as a list of definitions rather than an adaptation.

Finally, do not leave the video until the night before. A viva may follow, so you should be able to explain every choice without reading it.

What students ask about this kind of ECA

Search results and student forums show the same worries: how specific a prompt can be in 50 words, whether using GenAI will be penalised when it is allowed, how to show you did not just copy the output, and how to sound natural on camera. The answer to most is the same. Show your reasoning, label what came from the tool and change something visibly.

Frequently asked questions

Can I use ChatGPT or another tool for the LOG309 ECA?

The brief allows GenAI use with full attribution. You must state the tool, the prompt, the complete output and what you adapted. Check your course site for current rules, because a viva may be held and you must be able to explain your work.

Which technology should I choose for Q1?

Choose one that addresses the scenario and has a clear weakness you can fix in Q2. Demand sensing, a control tower or predictive arrival models all work. The best choice is the one you can explain clearly in about 200 words.

Does the API need working code?

No. The brief asks for a mini design document covering inputs, processing, outputs, integration and controls. A well-labelled table plus a short explanation of the link to your technology is the right format.

How do I pick the two constraints in Q3?

Pick the two your Q2 design already touches. Supplier data quality and maritime delays pair well because both involve uncertainty you can express as confidence scores and scenario ranges.

How long should the video really be?

It must run between 3 and 6 minutes. Plan for about 5 minutes so you have room for a slow start and still stay within the limit. Rehearse with a timer before recording.

Can I use a PowerPoint template for the slides?

No. The brief rules out templates from the internet. Build the deck from a blank file, keep the design simple and convert it to PDF with 15 slides or fewer.

Running short on time before the 30 October deadline? You can explore SUSS assignment help for model answers and feedback, check student reviews first, or find the SUSS question hub for related briefs. The service provides model answers and guidance for reference; how you use them under your university’s rules is your responsibility.

About the expert

Topics like GenAI prompting and API design fall within the specialisations of Wei Chen, who holds a Ph.D. in Computer Science from the University of Hong Kong and lists artificial intelligence, machine learning, data mining and natural language processing.

References

McAdoo, T. (2025, September 15). How to cite ChatGPT. APA Style. https://apastyle.apa.org/blog/how-to-cite-chatgpt

OpenAPI Initiative. (2021). OpenAPI Specification v3.1.0. https://spec.openapis.org/oas/v3.1.0.html

Singapore University of Social Sciences. (n.d.). LOG309 Industry 4.0 Logistics Applications. https://suss.edu.sg/courses/detail/log309