Choosing your tool for the job
Feature comparisons go out of date in weeks. A rule of thumb about the shape of the work lasts, because the shapes don't change as fast as the products.
Why we don't teach one tool
Most of the market teaches ChatGPT only, which quietly turns the training into a sales channel for one vendor. It also leaves people helpless the moment their workplace standardises on something else, which happens more often than anyone predicted.
Teaching two properly costs about twenty per cent more time and produces a skill that transfers. It also makes the data-handling conversation honest, because you can compare rather than assert.
The rule of thumb
| Shape of the work | Reach for |
|---|---|
| Long input, one careful answer — a bundle, a contract, a ruling, a plan set | Claude |
| Sustained drafting that has to sound like you across many documents | Claude |
| Reasoning through something with several moving parts | Claude |
| High volume, fast turnaround, twenty small jobs in an hour | ChatGPT |
| Talking instead of typing — voice note to finished draft | ChatGPT |
| Photo of a whiteboard, a form, a site, a label | ChatGPT |
| Anything where a wrong answer is a regulated harm, or identifiable records are required | Neither |
The longer version, with the reasoning, is on which tool for which job.
The part people skip
Both vendors treat your information differently depending on which tier you are on, and the free tiers are where most unofficial workplace use happens. That's not a footnote to tool choice, it's the main event for anyone holding client records — tiers and what happens to your data.
What you'd have at the end
- A one-line rule your team can apply without asking you.
- A short list of jobs in your business that belong to each tool.
- A written answer to the tier question for your business.
What this module covers
A rule of thumb for reaching for the right tool without thinking about it, built around the shape of the job rather than this month's feature list.
- Long, consequential, voice-sensitive work, where drift over a long input is the defect that matters.
- High-volume, fast-turnaround, spoken-in work, where friction is the thing that kills adoption.
- The third category: work that goes to neither, because a wrong output is a regulated harm rather than an inconvenience.
- Why stakes and length are two different axes, and what goes wrong when a team treats them as one.
What someone can do afterwards
Decides which tool to open without asking anyone, and can explain the decision to a colleague in a sentence. Teams without this rule default to whichever tool one confident person prefers, and the work quietly reshapes itself around that tool's weak side.
Where it sits in the order
Second or third. It needs the first module underneath it, because comparing tools is meaningless if neither is producing usable output yet.
Where this module stops
No benchmarks, no model version numbers, no feature tables. They date within a quarter and they aren't how the decision actually gets made. The longer argument is in the rule of thumb, and the tool-by-tool detail sits under tools.
The questions we get asked about this
Do we have to pay for both tools?
Usually not at first. One paid plan plus a free account on the other covers most small teams, and which one you pay for depends on which side of the split your work sits.
You don't resell either one?
No. No partner arrangement, no affiliate link, no referral fee in either direction, and where a free tier is genuinely sufficient we say so.
Get a training plan built for your business. Nine questions about how your business actually runs. You get a plan ordered for your work, a reading on where you're exposed, and one clear place to start. No account needed. Build your plan.
Start an enquiry