By Chisom Okafor- Obiudo
Two professionals ask ChatGPT the same question about investment risk in Namibia. One gets a generic textbook answer. The other gets a structured analysis they can take into a meeting. The difference is not the tool. It is what they typed into it.
Asking ChatGPT for a Namibian tourism pitch idea in Kunene region will produce an output that is broad in scope and surface-level in analysis. It may be informative, but it will lack strategic direction because the instruction itself lacks specificity
But if you assign it a role, a task, and a purpose, the output changes entirely. For example: “You are a sustainable development consultant. Create a luxury ecotourism investment pitch for a lodge in the Kunene Region targeting European conservation-minded travellers.”
The AI does not suddenly become smarter. Your instructions simply provide it with the map it needs to deliver output that is specific, relevant, and ready to work with.
Five Elements of an Effective AI Prompt
To avoid a garbage-in-garbage-out result, a good prompt includes the following five elements:
Start by assigning a role: “You are a senior credit analyst at a Namibian commercial bank.”
Next, provide context: ” Some clients in the fishing sector have applied for an expansion loan and have a history of late payments.”
Then state the specific task: “Summarise the key credit risks and recommend three conditions to attach to the approval.”
Add constraints: “The audience is a board risk committee. Use clear, non-technical language. Note that all names and figures in this prompt have already been anonymised for confidentiality.”
Finally, specify the format: “Structure the response as three short paragraphs covering risk summary, cash flow assessment, and recommended conditions.”
Notice how every element in the above prompt stems from your professional knowledge: your understanding of the audience, your awareness of confidentiality requirements, your judgment about what the output should include. None of that came from AI. That is what it means to apply your expertise when prompting. The framework simply gives your knowledge a structure that the AI tool can operate on.
One Line That Transforms Every Prompt
If the task is complex or you are unsure how to structure your request, add this instruction at the end of your prompt:
Before you proceed, ask me clarifying questions to ensure your answer is as relevant and accurate as possible
This changes the conversation. Instead of guessing your intent, the AI tool first asks you specific questions. You provide the missing information to ensure your outputs are sharper and closer to your requirements.
Example Prompts to Type into ChatGPT
Below are sample prompts for various Namibian sectors, fully anonymised. Notice how each assigns the AI tool an expert role and asks it to explain its reasoning.
Financial Services: Loan Risk Assessment
“You are a senior credit analyst at a commercial bank in Namibia. A client in the [specify sector] has applied for an expansion loan. Their debt-to-equity ratio is above the bank’s comfort threshold, their current ratio is below 1.0, and they have had late payments on an existing facility. Summarise the key credit risks, assess whether cash flow would support the repayment schedule, and recommend three conditions to attach to the approval. Explain your reasoning for each condition.”
Telecommunications: Board Member Meeting Preparation
“You are a non-executive board member of a Namibian telecommunications company. Management has proposed significant capital expenditure to expand 5G network coverage into several rural towns. Draft five strategic questions for the next board meeting. The questions must cover: (1) subscriber uptake projections, (2) expected monthly revenue per customer and affordability assumptions, (3) payback period and return metrics, (4) regulatory and permitting timelines/risks, and (5) alignment with Namibia’s National Broadband Policy (use the uploaded policy). Explain in 1–2 sentences why each question matters to the board. Avoid technical telecom jargon.”
Foreign Investment Promotion Strategy
“You are a senior investment promotion advisor at a Namibian Investment Promotion Agency. Identify three priority sectors for foreign direct investment, explain two competitive advantages Namibia offers in each, and recommend two actions to strengthen investor confidence within six months. Explain your reasoning for the sector selection. Before you respond, ask me clarifying questions so that your output is as relevant and accurate as possible.”
Press Release: Product Launch
“You are the head of corporate communications at a Namibian financial institution. The organisation has launched a digital lending platform that enables SMEs to apply for business loans online with a rapid turnaround. Draft a 250-word press release with a placeholder CEO quote emphasising financial inclusion. Use [Company Name] and [CEO Name] as placeholders. Explain your reasoning for the structure and messaging.”
Business Ideas for a Namibian Entrepreneur
“You are a business strategy consultant advising an entrepreneur in Windhoek with moderate start-up capital and logistics experience. Suggest five viable business ideas suited to the gaps in the Namibian market. For each, describe the target customer, a realistic capital range, one key risk, and what would make it succeed locally. Explain your recommendations. Before you respond, ask me clarifying questions so that your output is as relevant and accurate as possible.”
Five Common Prompting Mistakes to Avoid
Being too vague: “Write me a report on the Namibian economy,” tells the AI model nothing. You should specify the country, sector, period, and purpose.
Not assigning a role: “You are a senior credit analyst” produces a far more relevant and detailed response than a vague question about loan risk.
Accepting the first output: The first output should never be the final output. Read it with a critical eye, then give specific follow-up instructions such as:
‘Shorten this to 150 words.’ The tone is too informal for a regulatory submission. ‘Remove the jargon and explain the ratios in plain language.’ What other perspectives might I be overlooking? Or consider playing devil’s advocate to challenge my assumptions in this analysis.
Each instruction narrows the gap between what AI produced and what your professional standards demand.
Not asking the AI tool to explain its reasoning: Always add “Explain your reasoning.” This encourages AI to demonstrate its process, making it easier to identify flawed logic.
Not anonymising company-sensitive information: Never input confidential client names, real financial figures, or sensitive data into any AI tool. Use placeholders like Mr X instead of the Client Name, and general indicators instead of exact figures.
Each of these errors originates from the same core issue: excluding your expertise from the process. Reintroduce it, and the outcomes will speak for themselves.
AI does not threaten expertise; it exposes its absence. If you know your field, these tools enhance what you already have. If you don’t, then your priority should be to acquire that knowledge first. The quality of your thinking determines the quality of the output. Your expertise must remain in the operating system.
*Chisom Okafor-Obiudo is an admitted legal practitioner and Chief Legal Officer at the Namibian Law Reform and Development Commission. Beyond her work in law reform, she has developed a specialised expertise in corporate governance, AI governance, and AI literacy. She serves on the NCRST National Artificial Intelligence Technical Advisory Committee on law and governance. Reach her at chisomokafor11@gmail.com.
