AI for Data Analysis and Report Writing

Welcome and Programme Overview

Introduction to the session, objectives, and housekeeping.

Part 1: Introduction to AI and Generative AI What AI and Generative AI Actually Are

A plain-language explanation with no mathematics and no jargon. What a large language model is doing when it writes, and why that explains both its strengths and its failures. A guided tour of the three tools most Malaysian organisations already have access to: ChatGPT, Google Gemini and Microsoft Copilot. What each one is genuinely better at, and how to work out which is already licensed at your workplace.

What AI Can and Cannot Do

The honest capability boundary, drawn with examples rather than warnings. Hallucinations: why a model invents a confident, wrong answer, what that looks like in practice, and the verification habits that catch it before it reaches your manager. Confidentiality and what should never be pasted into a public tool. Where accountability stays with the human no matter what the tool produced.

Break

Part 2: Prompt Engineering Giving Clear and Effective Instructions

The anatomy of a prompt that works: role, task, context, output format and constraints. Why vague questions produce vague answers, and how a few extra sentences of context change the result entirely. Working through weak prompts and rewriting them live so participants can see the difference in the output rather than take it on trust.

Structuring Prompts and Practical Exercises

Hands-on prompting clinic using participants' own recurring work tasks. Iterating on a response instead of accepting the first draft, asking the model to critique its own output, and using examples to lock in the format you want. Participants start building a personal prompt library they keep after the training.

Part 3: AI for Data Analysis Understanding and Analysing Data with AI

Working with a spreadsheet directly in an AI tool. Cleaning and structuring messy data before analysis, which is where most attempts quietly fail. Asking questions that produce useful answers: summaries, comparisons, trends over time, and outliers worth investigating. Critically, how to check the arithmetic, because data work is where AI is most likely to be confidently wrong.

Identifying Insights, Trends and Key Findings

Moving from what the data says to what it means. Using AI to surface patterns you might not have looked for, then testing whether they hold up. Separating a genuine finding from a coincidence. Turning each finding into a plain-language statement that a non-analyst can read once and act on.

Lunch

Part 4: AI-Assisted Report Writing Structuring and Drafting Reports

Getting a usable outline before writing a single paragraph, and why this single step saves the most time. Drafting sections from your findings, summarising long source documents without losing the important qualifications, and writing an executive summary that genuinely summarises rather than restating the introduction.

Improving Clarity and Presentation

Editing AI output so it does not read like AI output: cutting padding, fixing the repetitive rhythm, and restoring your organisation's tone. Adjusting the same report for different readers, from an operational team to a board. Practical formatting for readability.

End-to-End Exercise

The final session puts the whole day together. Participants take a raw dataset through cleaning, analysis, findings and drafting to a finished short report in one sitting, using their own prompt library. This is the session that converts a day of demonstrations into a workflow people actually repeat on Monday.

Break

Who Should Attend:

Executives, officers and managers who prepare reports from data as part of their role. Finance, operations, HR, marketing and administrative staff who work with spreadsheets and written reporting. Government and GLC staff producing periodic reports and submissions. Anyone with limited or no prior AI experience who has been told to start using it and does not know where to begin. No technical or programming background is required.

Key Outcomes:

Explain in plain language what generative AI is and where its limits are Choose confidently between ChatGPT, Gemini and Copilot for a given task Recognise hallucinations and apply a verification routine before using AI output Write structured prompts that produce usable results on the first or second attempt Prepare and analyse a dataset with AI, including checking its arithmetic Turn raw data into clear findings a non-analyst can act on Draft, summarise and edit a report with AI assistance Leave with a personal prompt library and a completed sample report

Wrap-Up and Q&A

Key takeaways, next steps, and close.