AI in Action: Digesting Documents
Or how AI saved me a boring evening of work.
This week, I thought I’d share a smaller, quick cases where I’ve found AI especially helpful. Feel free to share your own in the comments or message me directly!
One of the keys for getting comfortable with AI is simply remembering to use it. It’s simple and sounds easy but it’s true. Whenever you tackle a task, keep this question in the back of your mind – Can I do this better, faster, or more easily with AI?
Here’s a recent example. I was collaborating on a project this week where we had over 20 related documents – email correspondence, professional reports, visual diagrams, etc. – and we needed to pull together a concise overview. One team member had set aside an evening to comb through everything. But wait – AI to the rescue!
This is a perfect scenario for AI, which excels at scanning and summarizing large volumes of information quickly.
To get started, we uploaded the documents into a Google Drive shared folder. Late last year Google introduced a Summarize this folder feature that makes interacting with your documents easy. Just by clicking the below button, Gemini opens an interactive side chat on the right and provided a high-level paragraph summary and three core bullet points that captured the essence of the content in seconds.
It did an excellent job. This is bread and butter stuff for LLMs, after all. I copied the text it gave me into a new summary document. Easy.
Next, I wondered what else we could generate. How about a timeline to help readers understand the sequence of events? I prompted Gemini with:
📝Prompt: Provide a timeline of the information in bullet form. For each bullet, reference the corresponding documents that the information comes from.
Again, Gemini did a fantastic job of taking a bunch of disparate pieces of information in 20+ documents and creating a coherent - and as far as I can tell, accurate! - overview of events. This saved a ton of work.
That said, AI isn’t perfect. Initially, Gemini didn’t include references, but after a quick follow-up prompt reminding it to do so, Gemini rewrote the timeline and corrected that. I also noticed that one of the references was incomplete and asked it why, to which Gemini responded “I see that was incomplete! You were right to let me know. I’ve corrected it.” and it was, indeed, fixed. A good reminder of the need to always review AI work (just like you would with human work).
Lastly, I thought it would be a good idea to index documents for anyone that wanted to dive in deeper, so I asked Gemini for that:
📝Prompt: Create an exhaustive list of all documents in this folder along with a one-sentence summary of each.
Perfectly executed. In under 20 minutes of time, I had taken a ton of documents and content and created from it:
A high-level summary of all content
A detailed timeline with source references
An overview of every document for deeper exploration
Honestly, pretty great!
For fun, I compared the results of Gemini with GPT-4o as well. A few takeaways:
GPT-4o summarized the content better (in my opinion) but its timeline output wasn’t quite as strong—though still usable.
Even though GPT said it could handle an unzipped file, it didn’t actually read everything inside, so be sure to unzip folders first before uploading.
That said, GPT does allow folder uploads directly into chat, and it automatically pulls in the contents—super convenient.
Lies! All lies!
Was this the most advanced use of AI ever? No. But was it effective? Absolutely. It saved hours of tedious work, and I was able to draft a solid executive summary in under 20 minutes. A fantastic everyday use case for AI.
Have anything to share that you’re working on? Post it up below.
