I have spent considerable time testing the leading artificial-intelligence platforms through actual work—not demonstrations or isolated prompts. My use has included financial research, technical validation, website development, document management and occasional creative projects.
These are not recommendations for everyone. They are observations from one demanding workflow and a snapshot of where the products stand for me today.
Copilot
Copilot may be the biggest tease of the group. It can produce a great deal of language without necessarily producing a finished product. In my experience, it often stops at the most obvious version of an assignment rather than taking the additional steps required to complete it.
Its integrations are the exception. Connections to Google Drive and other services can be genuinely useful. I continue to use Copilot for those practical connections, but it is rarely where I begin a complicated project.
Gemini
Gemini performed well for me early on. Over time, however, I found it increasingly difficult to understand the subscription levels across the Google products I already use. Some upgrades felt overlapping, and the possibility of paying twice for adjacent capabilities made the overall proposition less attractive.
Its presence in Chrome also affects the experience. Between the visible Gemini prompt, unsolicited AI summaries and sponsored search results, it can be difficult to separate the tool from the surrounding Google ecosystem. For some users, that integration may be convenient. For me, it can feel more persistent than helpful.
Grok
Grok offers considerable range through a relatively simple subscription. Its versatility is its principal advantage.
I can use it for a short financial or technical check, then turn to the same platform for a longer research exercise or a creative project such as producing a humorous video. It handles very different assignments without requiring me to spend much time deciding which product, workspace or operating mode to select first.
ChatGPT
ChatGPT was initially a baseline application that I used when I was not working in one of the others. That has changed.
Its integrations and its ability to remain engaged with complicated, multistep assignments have made it substantially more useful to me. It is particularly effective when a project requires information to be gathered across several files, applications or data sources while preserving the broader context of the work.
The distinction is not merely the quality of an individual answer. It is the ability to continue working through a messy assignment until the different pieces begin to form a usable product.
Claude
Claude is the most recent platform I have added and therefore the one with which I have the least experience. Even on a lower-priced plan, however, it has been a capable workhorse.
It has helped me manage large document exercises, consolidate information and impose order on complicated collections of material. It offers many capabilities similar to ChatGPT, but the experience can feel less inclined to expand every conversation into a larger project.
I have no substantial criticism to offer yet. Claude has been a pleasant surprise, and I expect to understand its strengths better as I continue using it.
Where I have landed
I do not need five platforms performing the same role. I need a small number of tools that complement one another.
At present, Grok and ChatGPT are the platforms I use most consistently. Claude has emerged as a useful third partner, particularly for document-heavy work. Gemini remains capable but increasingly difficult for me to separate from the broader Google product experience. Copilot is useful primarily when one of its integrations solves a specific problem.
The only comparison that has proved meaningful to me is not which model produces the most impressive answer. It is which tool helps turn an actual assignment into finished work.