"I want to use AI in equity research but I don’t know where to start."

I get this question a lot, so I thought I’d share how I use AI and why I use different tools for different purposes.

One thing I’ve learned: no tool is perfect. You have to figure out the right use case for each one.

Today, my workflow looks something like this:

📍 ChatGPT to simplify complex business models, map value chains, and understand where a company sits within an industry.

It works best when the information is relatively stable and requires synthesis.

📍 Gemini is usually my first stop when I need management interviews, news articles, or information on unlisted companies.

Its connection with Google’s ecosystem makes web discovery much easier.

📍 Google NotebookLM for conducting primary research through company documents:

Annual reports, concalls, investor presentations, sector reports.

Everything goes into one place.

I can challenge my assumptions and trace every answer back to the original source. It helps in strengthening your thesis.

📍 Claude Terminal has probably been the biggest upgrade to my workflow this year.

It sits on top of my local files and connects with the tools I use every day:

👉 Firecrawl for scraping websites.

👉 Obsidian for reading and editing research files.

👉 Obsidian Clipper for saving articles directly into my research folders.

👉 Playwright for interacting with databases like screener, stockscans etc. through my own credentials.

👉 NotebookLM which helps claude interact with my Google Notebook on any company and help create a first draft for anything related to company.

All the data comes from verified sources so minimal chance of hallucination.

The result is simple.

Instead of moving between 5 different applications, I give one instruction and the information flows into a single workspace.

That changes the economics of research.

The time spent searching, downloading, organising, and moving information around starts shrinking.

The time spent thinking starts expanding.

And that is where analysts still create most of their edge.

AI has helped me collect information faster.

The real benefit comes from creating more time to think about what that information means.


This workflow is really a system designed to reduce time.

Most people restart from scratch every time they use AI.

Instead of focusing on prompts, focus on building systems that work repeatedly and compound over time.

If want to understand more on implementing this workflow, you can DM.

Follow Nikhil Singh to stay updated with more such interesting content and insights.