Okay, so I have to share this because it was a genuine "why didn't I do this sooner?" moment. My team's been managing a systematic literature review project, and our old process was... painful. We'd pull hundreds of PDFs into EndNote, and then two of us would spend days manually tagging them with custom keywords for themes, methodologies, and applicability scores. The inconsistency and sheer time sink were killing us.
We trialed Iris.ai for its filtering engine, but the real game-changer was using their **AI-generated research topics & categories** feature. Instead of us defining all the tags upfront, we uploaded our document set and let Iris.ai's engine analyze the full text. It automatically clustered the papers and suggested thematic categories.
Here's the concrete win:
* The AI surfaced a subtle methodological sub-theme we'd completely missed in our manual scan—about 15 papers we would have incorrectly deprioritized.
* It cut out the entire manual tagging debate loop ("Does this *really* fit under 'neural networks' or 'deep learning'?"). The AI's categories weren't perfect, but they were a 90% accurate starting point we could refine in minutes, not hours.
* We estimated the time saved just on the categorization and initial sort phase to be about **20 hours** for this batch of ~400 papers. That's huge for our timeline.
The workflow shift was key: we went from **Define -> Tag -> Analyze** to **Upload -> Analyze -> Refine**. It flipped the script. You do need to review the AI's suggestions—it sometimes makes odd splits—but that's far faster than building the structure from zero.
For anyone doing structured reviews or meta-analyses, this is the feature to look at. It feels less like a fancy search tool and more like an automated research assistant doing the first grueling pass. Has anyone else compared this to other AI-driven categorization tools like those in Elicit or even Zotero's new ML features? I'm curious about the consistency across platforms.
Spreadsheets > marketing slides.