For the last 2 years I have been studying for CFA and I am loving the curriculum and the prep. Although, with the onset of AI Tools, I felt job roles of Research Analyst or anything related to gathering data, refining it and presenting, would be taken over by AI. Considering the opportunities AI Tools provide in conducting Research efficiently, I decided to attent some AI trainings, as I am not from the science backgroud.
Last week, I attended an AI Masterclass. I was introduced to many tools. I experimented on them, but initially the output was not good as my prompts were very vague. So I decided to watch hours of Prompt Engineering videos on youtube and create a prompt, that will give me my desired results. After a lot of RnD, I finally got a satisfactory report. It is presentable, detailed and easy to understand. I am attaching the report for your views and comments, on how it can be improved. Feel free to share it with anyone who has a much better understanding in this.
@ishu - Thanks Ishan for this one. The report has come out really well. Definitely a good amount of efforts have been invested by you. If possible, please share from your learning curve, how we can generate this kind of a report. I am assuming this report is a mix of Quantitative results + Commentary from Earnings call. Keep up the good work of sharing these nuggets over.
hi Pratik. Actually I have not used any source files for generating the report. I tried to create a report using 2 ways.
NOTEBOOKLM
If you are a beginner, this tool is just awesome. You can add pdfs, excel sheet, website urls and even youtube videos as reference. You can then ask whatever you need but it can only source data from the sources you have attached for its reference. So when I created a report on Notebooklm, it was not satisfactory. The report was plain text, no layout, no charts, no tables, etc. This tool allows creating tables, it is not very impressive.
Perplexity, Grok & Claude
Now this 3 step process helps in going deeper in research and presentation.
Starting with Perplexity. This tool has a much better research capabilities than NotebookLm and Chatgpt. It can conduct independent research and get you data accross 100s of website, and give a detailed text report as an initial source for the next step.
Second is Grok. I use this tool to create the text that will be included in the final report, like company overview, revenue sources, headwinds, tailwinds and interesting facts. This tool also creates a step by step process for step 3.
Last step is Claude. This tool creates the final report in minutes. Once you add the output from Grok and feed it to Claude, it will follow the exact steps mentioned and make you a research report. This is a very short version of the process. It took me 2 hours to complete the report. It feels like it takes a lot of time, but if I tried making a similar report with the knowledge I have of excel, word, powerpoint and my researching skills, it will take at least 2 days at warp speed to create something similar.
This is not the perfect version, as there is always room for improvement. This is a start. So I will keep sharing more such reports, trying to improve it every single time.
I have been trying my hand on various AI tool off late for equity research and no doubt Notebook LLM is best for getting 1st hand information from company’s Annual report and earning calls.
I have tried perplexity, Chat GPT, Claude and Gemini all for external research like peers and industry analysis
Though perplexity is very good, but recent Deep Research option of Gemini is really give detailed reports. Attaching one of the report of Aditya Birla real estate sample extracted from one single prompt and Gemini gave all the relevant data to be used for research.
Though I am finding it difficult to summarise and format the report using AI in desired presentatable tabular format with visually appealing look, Ishu you did wonderful job in this.
If you can share link of YouTube videos that helped you to summarise with claude would be really helpful for everyone genuinely interested in learning AI for reporting.[quote=“ishu, post:1, topic:197969, full:true”]
No I did not use paid versions for this. Although I would suggest to subscribe for better services if you want to use the tools for extended period of time. Sometimes, due to daily limits and memory restrictions, the process is very slow. I am sure, paid versions will generate much better reports. But for day to day use, free versions are good.
Really appreciate your initiative here — the report looks great for someone who’s just getting into AI tools with a finance background.
I’m the founder of CandleYogi.com, where we’re building tools to simplify the research and analysis workflow for retail traders. Your CFA background complements my tech side quite well, and I was thinking — why don’t we combine forces?
One insight I’ve noticed: a lot of well-structured research is just built on publicly available data — yet it sits behind expensive paywalls, making it inaccessible for most retail traders. If we can automate and open-source this process, we can actually democratize financial research for a much wider audience.
Let me know if you’re open to a quick chat — I think there’s real potential here.
@ishu thanks, your guidance helped me to use Claude AI for better reporting format, I was able to bulid template to automate my pre-designed format, though few details and focus of report needs to be re-aligned with hit and trial but progress is really impressive I am attaching sample report (though Original report was done in HTML format which I feel is good format to view across all decvice, mobile, tablet and Laptop) so printing to PDF has some formatting issue… I thought sharing this format may be helpful for some eMudhra Limited - Equity Research Report.pdf (1.0 MB)
Once again thanks for guiding to rightful direction
Hi Forum,
I am a final-year MBA student from the Delhi School of Economics. I have been interested in the stock market for the past few years, especially in understanding business models and how companies create and sustain value.
One thing I’ve come to realise is that great investing is not just about numbers, but about developing a deep understanding of businesses something I find many members of this forum are incredibly good at. I’ve been a long-time reader here, and I genuinely admire the quality of analysis that goes into the threads. In fact, some of your reports don’t just read like analysis they feel like you’ve put your soul into them.
Inspired by this, I’ve started working on identifying small-cap and mid-cap companies with differentiated or niche business models. I’m also experimenting with AI tools not to shortcut the process, but to make my research deeper and more structured. I’m trying to use AI as an assistant while ensuring that the final output reflects real human thinking, effort, and curiosity.
In this post, I’m sharing my research report a company I found quite interesting because of its Maxposure Limited Reinventing the Aviation Media-Tech Ecosystem.pdf (496.6 KB)
I’d be extremely grateful if some of you could take a look and share your thoughts. I’m still learning both about investing and about improving one’s thinking, and input from more seasoned participants like you would be very helpful. I’m looking forward to learning and contributing.ng more in this amazing forum.
Great Report man. One tip that has helped me lot is giving a detailed context in the prompt.
My prompt looked something like this-
I am an equity research analyst at a company similar to JP Morgan. I am working on a very important equity research report on Nuvama Wealth. The report needs to have a professional layout. You can also use graphs and charts to explain the data in detail.
This is a very basic prompt but it conveys the point ig. The context being “I am an Equity Analyst at a company similar to JP Morgan.” This compels the tool to think and look for JP Morgan reports, which gives the report a professional layout, without you having to explain it in detail.
Anyone with knowledge of how to link Google sheet or CSV data as source in Notebook LLM?
I export data from Screener.in with one click I just need that data to be used by AI this save lot of time and token as ai will not have to research financial data from annual report and investor presentation and than format back to my desired table. This considerably reduce size of prompt as well .
I have tried Script to link Google sheet data to Google docs which can be used as source for notebook LLM but there are some authentication issue.
Any body have tried their hands on connecting google sheet with notebook LLM?
Yes I have tried screener url before, as I understand it uses default column so additional information for peer comparison etc. is not fetched completely, csv file that I export has much more data colum.
in Google drive option of notebook LLM only docs, ppt can be linked, no way to link google sheet or csv
You should consider Perplexity. It can do a deep internet research on its own. If you give it a detailed task and don’t add any attachments, it will surf the internet to get the information.
