AI for Research & Analysis
6.2 — AI for Research and Competitive Analysis
What You'll Learn
Research has always been time-consuming. Gathering information, reading sources, pulling out the relevant bits, seeing patterns across dozens of documents — it's work that used to take days. AI has genuinely transformed this part of professional life, and this lesson will show you exactly how to use it well.
You'll also learn where AI falls short in research — so you don't walk into a meeting quoting something that was made up.
Where AI Genuinely Saves You Time
Let's start with the honest wins.
Summarising long documents. Paste a 30-page report into Claude or ChatGPT and ask it to pull out the five most important findings. This takes seconds, not hours. The same works for academic papers, legal documents, earnings reports, policy documents — anything dense with text.
Synthesising across multiple sources. Once you've gathered your sources, AI can help you see what they have in common, where they disagree, and what the overall picture looks like. This is pattern recognition at scale.
Generating research questions. If you're not sure where to start, AI can help you identify the right questions to investigate. "I'm researching how our competitors are positioning their pricing — what should I be trying to find out?" is a legitimate and useful prompt.
Competitive landscape overviews. AI has broad knowledge of industries, companies, and markets up to its training cutoff. For a general overview of how a market works or who the major players are, it's a useful starting point.
Turning raw notes into structured analysis. You've done the interviews or read the articles. You have a pile of notes. AI can help organise those notes into a coherent structure — grouping themes, spotting contradictions, identifying gaps.
A Practical Workflow for Research Projects
Here's a workflow that combines AI with human judgment effectively.
Step 1: Define your research question clearly. Before you involve AI at all, write down exactly what you're trying to find out. Vague research produces vague answers. "What's happening in the CRM market?" is worse than "What are the three most common reasons mid-size companies switch CRM providers, and which tools do they typically move to?"
Step 2: Use AI to generate a research framework. Ask: "I need to conduct competitive analysis on [industry/product]. What aspects should I research, and what sources should I look at?" This gives you a structure before you start searching.
Step 3: Gather your sources — yourself. This is important. For current, reliable research, you need to find the actual sources: industry reports, company websites, news articles, analyst commentary. AI's training data has a cutoff date and it cannot browse the internet reliably unless you're using a tool specifically set up for that (like Perplexity AI or ChatGPT with web browsing enabled).
Step 4: Feed the sources to AI for synthesis. Paste in your gathered material and ask AI to analyse it. "Here are excerpts from three competitor websites and two analyst reports. Summarise how each competitor is positioning themselves on pricing, and identify any gaps in the market." This is where AI earns its keep.
Step 5: Human review and judgment. Read the synthesis. Ask: does this match what I know? Is there anything surprising that needs verification? Add your own interpretation — context that AI doesn't have about your company, your customers, your market.
Key takeaway: Think of AI as a very fast, very well-read research assistant who needs clear direction and human review. It can do the reading; you need to do the thinking.
Generating Insights from Data
If you have structured data — a spreadsheet of sales figures, a set of survey responses, a table of competitor features — AI can help you make sense of it.
Example prompt with data:
"Here is a table of our customer satisfaction scores across 5 product lines for the past 3 quarters. What patterns do you see? Where are scores improving, declining, or flat? What might be worth investigating further?"
Paste in the data as text (or use ChatGPT's data analysis feature, covered in lesson 6.3), and AI can give you a useful first-pass analysis in seconds.
What it can't do is tell you why the patterns exist, or what you should do about them. That requires human judgment, domain knowledge, and often more research.
Where You Still Need Humans
Be clear-eyed about the limits.
Current events and real-time data. AI's knowledge has a cutoff date. For anything that happened in the last few months, verify with current sources.
Primary research. AI cannot interview customers, run surveys, or attend industry conferences on your behalf. Original insights still come from people.
Verification of specific facts and figures. AI can and does hallucinate — it will confidently state a statistic that doesn't exist. Any specific claim that matters should be verified against a primary source.
Strategic judgment. AI can describe what competitors are doing. It cannot tell you what your company should do about it. That requires understanding your resources, your culture, your customers, and your opportunities — none of which AI has access to.
Sensitive or proprietary research. Don't paste confidential company information, client data, or trade secrets into a public AI tool. More on this in lesson 6.6.
Practical Tips
- Use Perplexity AI for research that needs current sources. Unlike base ChatGPT, Perplexity is designed to search the web and cite sources — far more reliable for recent information.
- Ask AI to identify what's missing. "Based on this analysis, what questions can't we answer yet, and what information would we need to answer them?" is a powerful prompt.
- Ask for multiple perspectives. "What would a sceptic say about this conclusion?" helps stress-test your analysis.
- Build a document as you go. Paste AI synthesis outputs into a running document. Treat them as input to your thinking, not the final output.
Key takeaway: AI is transformatively useful for the mechanical parts of research — reading, summarising, pattern-spotting. The intellectual parts — judgment, strategy, verification — still belong to you.
What to Try This Week
Take a research task on your plate — a competitive review, a market overview, an analysis of customer feedback. Try using AI to summarise your sources after you've gathered them, and ask it to identify patterns. Then compare what it gives you to what you'd have written yourself. Notice what it gets right, what it misses, and where it adds genuine value.