"Claude or ChatGPT?" only has a useful answer in front of a concrete piece of work. Testing a second tool makes sense when there is a real question: reviewing long documents, holding a writing standard or working better with a project's materials. That question should drive the comparison, not this week's ranking.
Start with a deliverable someone on the team knows well. If no one can recognize an error, a convincing answer may seem good enough even when it changes the meaning of the material.
Set up a fair comparison between Claude and ChatGPT
Use the same set of approved documents and an equivalent request. Record which features were available in each account: a difference in plan, settings or access can shape the result as much as the model.
When reviewing an internal report, the task might be to find inconsistencies between the executive summary and the data, citing the source of each observation. What matters is how faithful the analysis is, how easy it is to check the findings and how much work it takes to fix the deliverable.
Run a second round with one change: an updated number or a new constraint. Watch how each tool incorporates the revision. That continuity matters for work that goes through several versions.
Look at the environment along with the answer
In Claude, Projects let you bring together context and documents, and Artifacts offer a space for editable deliverables. They are features to explore within the chosen task; availability needs to be checked in your account.
Note what was finished, what needed correction and what was out of reach. If one tool delivered a clearer analysis but made the review harder, that trade-off has to show up in the evaluation.
The right choice is one you can explain
At the end, the team should be able to say when it would use each option — or conclude that the current setup already does the job. The value of the test is in building your own criteria, which you can revisit when the work or the features change.
Tools change with every launch; the criteria stay. That is the starting point of AI Thinking, Dani Hayoshi's method: before choosing the tool, decide what the work needs to make possible and who does each step. To organize several tools at once, see the AI tool map built from the company's work.
Dani Hayoshi is the author of AI Thinking. Her work connects artificial intelligence, education and entrepreneurship. About Dani.