You can get a finished, cited research report from three different AIs now, and the right one depends on one thing: where your files already live.
Deep research isn't new, what's new is what comes out the other end. On September 9 both OpenAI and Google shipped versions that hand you a document, not a chat reply. Claude has done that in Cowork for a while. So the question moved from "which one researches best" to "which one lands the report where I'll actually use it".
The three, in one line each
ChatGPT Deep Research, in Work mode. Searches the web, your uploaded files and connected apps, and returns an editable document with citations. Plus and up, Work mode, uses your Work allowance. Since September 10 the Library also reaches into Dropbox, Box and SharePoint, so if your files live there, this is the one.
Gemini Deep Research, in the Docs side panel. Pulls across Drive folders and Gmail threads plus the web into a report with sources, saved straight into Drive. Google AI Pro and Ultra. If your whole life is in Google, stop reading and use this.
Claude Research, in chat and Cowork. Web plus whatever you've connected (Gmail, Drive, Notion and so on), with citations. In Cowork the report lands as a file in the folder you pointed it at, next to the material it was built from. Paid plans. If your files live in folders on your computer, or in Notion, it's this one.
Now a quick word from today's sponsor.
The Future of AI in Marketing. Your Shortcut to Smarter, Faster Marketing.
This guide distills 10 AI strategies from industry leaders that are transforming marketing.
Learn how HubSpot's engineering team achieved 15-20% productivity gains with AI
Learn how AI-driven emails achieved 94% higher conversion rates
Discover 7 ways to enhance your marketing strategy with AI.
Right. Back to it.
Where your files live decides it
Google Drive: Gemini. Dropbox, Box or SharePoint: ChatGPT. A folder on your computer, or Notion: Claude. Files everywhere: ChatGPT, because its Library is the widest net.
I think the quality gap between the three on a normal business brief is smaller than the gap between a good brief and a bad one. The report is won in the prompt.
The brief that works in all three
Research brief: [one sentence on the decision this report has to support].
Sources: use the web and [my Drive folder / my uploaded files / the connected apps]. Prefer the last 12 months. Cite every claim with a link or a file name.
Output: a document, not a chat answer. Sections: what we know, what's disputed, what's missing, recommendation. Two pages maximum. If two sources disagree, show both and say which you trust and why.
Do not pad. If the answer is short, the document is short.Run it on a real decision you have this week, in whichever tool your files point you to. Read "what's missing" first. That's the section where it stops pretending.
Where I'd be careful
A citation means it found something, not that the something is right. A lawyer in New Mexico was fined $5,000 this month for an AI-written appeal with invented witnesses, and that filing had citations too. Click three of them. If all three hold, trust the rest. If one doesn't, redo the brief.
Make it stick
Keep the brief as a template in the tool you picked: a Project in ChatGPT, a Gem in Gemini, a Project in Claude. Then a research report is a one-line request with the decision filled in.
That's one report you don't spend Tuesday night on. An owner's month has a dozen decisions that deserved one.
If you own a business and want the boring work off your plate, that's what Run On Claude is for. Every repeat job you'd hire for, set up on Claude so it runs on autopilot, with the install files. You keep the decisions. Owners only. $147 a month until member 50.
Quick one.
What decision are you sitting on because the research feels like a week? Hit reply.
Talk soon,
Zephyr
