ChatGPT gets most of the headlines. It was the model that made AI mainstream in late 2022. It's the one your clients, staff, and family are using. And it's genuinely good at certain tasks that Claude and Gemini handle less well.
This module is about those tasks specifically, so you know when to reach for it rather than sticking with whichever model you happened to open first.
---
What ChatGPT Does Particularly Well
Image input. On the ChatGPT Plus tier (and on the free tier in some instances), you can upload photos and ask ChatGPT to reason about them. For small business owners, this is immediately practical.
A tradie can photograph a job site, upload it, and ask: "Based on this photo, what materials am I looking at for this bathroom renovation? Give me a rough list." It won't quote you prices (see Module 3 on hallucinations), but it'll give you a starting materials list faster than staring at the photo for ten minutes.
A retailer can photograph a product, upload it, and ask for three Instagram captions, a product description, and a suggested category for their website.
A property manager can photograph a maintenance issue and ask for an initial assessment of what trade is needed.
Image understanding is a genuine advantage for trades and hospitality businesses where so much of the work is visual.
Code interpreter (data analysis in a CSV). On ChatGPT Plus, you can upload a spreadsheet, a sales report, your bookings data, a stock inventory, and ask it to analyse it. It will run actual code on the file, not just read it. Ask it to summarise your best-selling products, find the months with the highest and lowest revenue, or flag any data anomalies.
For business owners without a data analyst, this is genuinely powerful. You can upload a year of transaction data and ask: "Which product category had the highest revenue? Which had the highest return rate? Show me month-on-month trends." It'll produce a real analysis, often with charts.
Custom GPTs. OpenAI has a library of user-built "Custom GPTs", specialised versions of ChatGPT pre-tuned for specific tasks. There are Custom GPTs built for specific industries: legal drafting, HR policy, content creation for real estate. Some are good, some are mediocre. Worth browsing if you have a specific repetitive task, someone may have already built a decent tool.
Voice mode. ChatGPT's voice mode (on the mobile app) allows real conversation. You speak, it responds, you speak again. For thinking out loud, drafting while driving (please use hands-free), or quickly capturing ideas, this is useful. Claude and Gemini have voice features too, but ChatGPT's voice mode has been more polished in practical use for longer.
---
A Real Business Use Case: Tradie Photo-to-Quote
Scenario: a plumber attending a job needs a rough scope of works to quote from.
Steps:
- Open ChatGPT on mobile.
- Take a photo of the area of work (e.g. under a bathroom vanity showing the existing pipe configuration).
- Upload it in the chat.
- Type: "I'm a plumber quoting on replacing the vanity in this bathroom. Based on this photo, what work is likely involved? List the tasks I'd need to include in a quote."
ChatGPT will look at the image and produce a list of likely scope items. It won't know your prices or whether there's asbestos in the wall, but it gives you a starting checklist in 60 seconds that you then verify on-site.
This isn't replacing your quoting expertise. It's reducing the cognitive load of generating a first-pass checklist from memory.
---
Another Real Use Case: Sales Data in a Spreadsheet
A small café owner exports her weekly sales report from her POS system as a CSV. It has columns for date, item, quantity, and revenue.
She uploads it into ChatGPT Plus's code interpreter.
She asks: "Which five items generated the most revenue last week? Which five sold the most units? Are there any items that sold a lot of units but generated almost no revenue (potential pricing issues)?"
ChatGPT runs actual Python code on the file, produces a ranked list, and flags two items that are popular but priced below comparable competitors based on the unit economics it can see in the data.
Twenty minutes of analysis, done in four minutes. She adjusts the price of one item the following Monday.
---
Where ChatGPT Falls Short
Verbose hedging on complex questions. Ask ChatGPT something ambiguous or ethically complex and it will often produce an essay of caveats before getting to the answer. Claude tends to be more direct on nuanced reasoning tasks.
Sycophancy. This is a real, documented behaviour in ChatGPT. If you tell it your idea is good before asking for feedback, it will be less likely to give you critical feedback. If you push back on its critique, it will sometimes soften the critique to agree with you. This is called sycophancy and it's a known limitation.
The fix: actively ask for the opposite. "What's wrong with this plan?" rather than "Is this a good plan?" And when it pushes back, ask "If you had to argue the other side, what would you say?"
Cost adds up on Plus. The paid tier is real money for a business watching costs. If you're using it heavily (code interpreter, image analysis, voice mode), it pays for itself quickly. If you only need basic text generation, the free tier does most of what you need. OpenAI publishes current pricing on their site; check before you commit.
---