Part 3 · Working With AI Tools
AI-Powered Analysis Without Coding
The Data Is Already on Your Desk
Deep Research sent AI across many sources to build one grounded answer. Now the data is already in front of you: a spreadsheet three hundred rows deep, and a question you cannot answer by scrolling: what changed last quarter? For years, answering that meant a formula, a pivot table, or a favour from the analytics team.
Why This Matters
The gap between having data and understanding it costs most professionals hours every week, and it keeps the answers to your own questions sitting in someone else's queue. Close it, and analysis becomes a conversation you run yourself. By the end of this lesson you will be able to:
- Perform a trend summary on a real dataset in plain language
- Compare options side by side without building a model
- Extract the key points from a long document in minutes
This is Data Proficiency: reasoning about data no longer requires code.
Analysis is now a conversation
You are not writing code that runs on your data. You give the data to an AI assistant, and it reads and reasons over it in language, the way a capable analyst would, but in seconds. What separates a useful answer from a misleading one is not technical. It is precision: being exact about what you are asking, and refusing to trust the result until you have checked it.
- Provide the data: upload the file, paste the table, or point to the document. The AI reads it directly.
- Ask in plain language: name the question and what a good answer looks like, which three products declined, and by how much?
- Check the working: ask it to show the figures it used, then verify them against your source. AI can miscount or invent a trend.
That third move is not optional. AI can transpose a number, miscount rows, or describe a pattern that is not there, and do it confidently. For anything that informs a decision, you remain the analyst of record: the AI drafts the analysis, you verify it and you own the conclusion.
Key Takeaway
The AI drafts the analysis; you verify it and own the conclusion.
Check your understanding
Self-check only. Nothing is scored and nothing is recorded; open an answer to compare it with your own.
A procurement officer has a spreadsheet with 18 months of supplier delivery data (dates, quantities, defect counts, and costs) for five suppliers. She wants to recommend which supplier to renew the contract with. She uploads the spreadsheet to an AI tool and types: 'Which supplier is best?' The AI gives a vague, non-committal answer. What is the most effective way to improve the result?
- Upload the spreadsheet to a different AI tool, this one cannot handle data analysis.
- Rewrite the prompt with the three moves from this lesson: provide the data (the upload), ask in plain language (compare the five suppliers on on-time delivery, defects, and cost per unit, and recommend which to renew), and check the working (ask it to show the figures it used, then verify them against the spreadsheet).
- Add more data; 18 months is not enough for AI to find patterns.
- Ask the AI to write a Python script to analyse the data, then run the script.
Show the answer
Rewrite the prompt with the three moves from this lesson: provide the data (the upload), ask in plain language (compare the five suppliers on on-time delivery, defects, and cost per unit, and recommend which to renew), and check the working (ask it to show the figures it used, then verify them against the spreadsheet).
A project manager has three documents: a 20-page project status report, a spreadsheet of task completion dates, and a folder of client email feedback. She wants to understand why the project is behind schedule. She uploads all three to an AI tool and asks: 'Analyse these documents and tell me what is going on.' The AI returns a generic summary that restates information she already knows. What should she do differently?
- Upload only the spreadsheet, the other documents are too unstructured for analysis.
- Prepare each document (clean, structure, format), label them clearly, and rewrite the prompt: 'Source 1 is the Q2 project status report. Source 2 is the task completion timeline. Source 3 is client feedback emails. Compare planned vs actual timelines in Source 2, identify the root causes mentioned in Sources 1 and 3, and list the top three factors causing delays with evidence from all three sources.'
- Ask three separate questions (one per document) then combine the answers manually.
- The AI output is fine, she should accept the summary and move on.
Show the answer
Prepare each document (clean, structure, format), label them clearly, and rewrite the prompt: 'Source 1 is the Q2 project status report. Source 2 is the task completion timeline. Source 3 is client feedback emails. Compare planned vs actual timelines in Source 2, identify the root causes mentioned in Sources 1 and 3, and list the top three factors causing delays with evidence from all three sources.'
Saved in this browser only. Carry progress between devices with the resume link on the course page.