Before the assignment, my process depended heavily on normal ChatGPT or Claude AI. I would take whatever they gave me, check the sources one by one, and hope the information matched. It worked “okay”, but looking back, most of the answers did not actually appear anywhere in the cited sources. That turned every assignment into a long, uncertain cycle of verifying, correcting, and rewriting. One example that still sticks with me was a group project on FDI inflows of Singapore. My teammates barely did anything, so the entire research load fell on me. I relied on ChatGPT to speed things up, but it gave me almost all incorrect or irrelevant sources, only two out of everything it generated were actually valid. I ended up spending seven hours straight reviewing every citation, replacing broken sources, and generating new prompts just to get enough accurate material to finish the paper. That experience showed me how unreliable and time-consuming chatgbt normal method actually was. The dot-com vs. AI bubble project made this weakness even more obvious. Normal ChatGPT produced unlinkable citations again and again. 

Rabbit disrupted that completely. It works like a search-engine map for research: I put in a topic, and it instantly shows me clusters of real studies, how they connect, and which ones are closest to my question. When I looked up the relationship between the dot-com boom and the current AI boom, I expected confirmation that the AI bubble will repeat what happened in the dot-com bubble in terms of extreme growth and then recession caused by investors investing in growth companies in dot-com era and these companies failed leading to people loosing money which caused recssion (this was my initial thesis). Instead, Rabbit led me to papers that argued the opposite, that the AI boom is structurally different, with stronger fundamentals, real revenue, and wider adoption. It didn’t just “give me an answer”; it forced me to see where my assumption might be incomplete.

That was the part I appreciated the most: Rabbit didn’t fabricate anything. It specialized, filtered, and guided me toward papers I would never have found on my own. It helped me build a more accurate analysis of technological bubbles and understand why the dot-com crash unfolded the way it did, inflated valuations, lack of real business models, and companies going public based on hype rather than fundamentals. Having access to real research made the historical comparison stronger and more grounded. But Rabbit also revealed its limitations. It isn’t a summarizer. It doesn’t highlight the key arguments for you. It doesn’t save time in the sense of “read less”. Instead, it saves time by helping you find the right papers, but you still have to sit with it, analyze them, and extract what matters. It’s a powerful discovery tool, but a very hands-on one.

Deep Research chatgbt, on the other hand, approached the same topic from the opposite direction (I asked this time on case study on AI bubble and how it will be close to dot-com bubble companies). Rather than mapping the literature, it synthesized it. I liked how it asked me clarifying questions, for example, whether I wanted a case study on a specific company like Nvidia or Amazon, because it made the final output more tailored to my exact research question. And unlike normal ChatGPT, the links actually worked, the sources were real, and the structure felt much closer to a mini-review paper. It definitely saved time because it summarized and organized the main findings from multiple sources into something coherent.

But chatgbt Deep Research also had its drawbacks. It loaded information very slowly and took time to find the sources. And occasionally it pulled older sources when I needed the most recent data like “Investors and regulators have learned from past bubbles. After the dot-com crash, laws like Sarbanes–Oxley (2002)” which I feel may find a more accurate and recent source for it. It still felt far more reliable than regular ChatGPT, but not perfect.

Comparing these tools to how I used to do research makes the difference very clear. Before, I was relying on intuition and trial-and-error, whatever seemed relevant, I grabbed, even if I couldn’t confirm its reliability. Now, my workflow feels more professional and more academically grounded. Rabbit gives me accuracy; Deep Research gives me clarity, depth, and saved me time on reading all sources. One builds the map, the other tells the story. And using both gave me a stronger understanding of technological bubbles than anything I could’ve done with my old process. I found myself preferring ChatGPT Deep Research. Rabbit is incredible for discovering the right papers, but Deep Research actually shapes those papers into something usable. It summarizes, filters, and organizes the information in a way that fits the exact angle I care about. Instead of spending hours reading every article, Deep Research gives me a structured interpretation. It also interacts with me, asking for clarifications, adjusting the direction, and tailoring the output until it matches my research goals like for instance  it asked me what type of case study I exactly want for my research, big companies like Nvida and Microsoft or other fintech companies which was really cool. That level of adaptability makes the whole process much faster and more targeted.

This experience didn’t just improve my paper. It changed the way I think about research itself, the balance between discovery and synthesis, and the importance of tools that keep me both honest and efficient.

https://chatgpt.com/share/693436e8-9094-800c-9d25-9a13edae72a5 link with chatgbt deep research https://drive.google.com/drive/folders/1RfJPU6WSmaxQh7SJR6OxAuNH9xjlh5T5?usp=sharing link with the photo use https://app.researchrabbit.ai/library?folders=typefolderid7ebcab77-fc8f-4a9e-ae18-bdcef3d8146c i dont know how i can link rabbit prompt