Deep Research Topics for the Rest of Us

Deep Research is a service now offered by several AI providers like Gemini, OpenAI, Perplexity, and now Manus. It leverages agentic AI that performs an extensive research by visiting websites or even fetching data from APIs instead of simply answering your question by whatever knowledge is embedded in the Large Language Model itself. The result takes longer to generate (I have seen it take up to an hour) but is a more extensive document that would have probably taken several hours of research if you wanted to generate it yourself. I recommend anyone to try it. It is impressive.

OpenAI Deep Research now has a limited general availability. You can initiate up to 10 deep researches monthly if you are a paying user. Also, Manus from China was released, and it seemingly has more permissive limitations than OpenAI. It is only harder to get an invite. However, one of my bigger challenges with these tools isn’t getting access, but knowing what to use them for. It turns out that I don’t have too many open research questions that I can pose to these tools. It is disappointing to be given such strong tools but not knowing what to use them for. Here are a few ways that I have used deep research in my life, and I would love to find out about more ways that I can use it.

Investment Analysis

I have written a full-context about where I am in life and my investment goals. I have then also provided a breakdown of my portfolio holdings in percentages. Then, I have asked Deep Research to:

• Derive an investment thesis based on the distribution of my assets.
• Make suggestions for potential improvements based on my inferred investment thesis.
• Describe likely scenarios where my portfolio thrives and where it suffers.
• What can I expect from my portfolio given the current macro conditions?
• What improvements could I be making?
• And what are some other public investors with whom my investing style overlaps?

Essentially, I am providing context to understand my revealed preferences for self-discovery and looking for ways to optimize my current practices.

Results

I loved the results generated by both OpenAI and Manus. They helped me understand my implicit investment preferences and optimization methods and discover other sources, like public investors or books, that align with my way of thinking. You can easily ask for other investment approaches that would challenge your current ways of doing things as well.

Optimizing Health

I have been taking 300mg of Magnesium supplement before going to bed for a while now and it has been one of the best adjustments in my life. I used to wake up frequently before using these supplements, however, that is not the case anymore. Given the improvement that a single supplement had in my life, I have initiated a research on what other things I could use to improve my life quality. I have provided my full health context to Deep Research and I have asked it to:

• Provide more recommendations for new supplements to use, given my health goals, sorted by the likelihood that they can help
• Include any potential side-effects that I should be aware of
• Give a detailed explanation of how to integrate those supplements into my life, and work them into my schedule.

Results

This was another research topic that AI did a really good job generating. One of the best things about getting this information from AI is how it can customize your intake based on your schedule and other inputs. The worst part is that the information could be faulty and needs to be double-checked. OpenAI deep research told me that I should be careful when taking Ashwaganda if I am planning on getting pregnant (while referring to me as a male). The research report includes links so you can use the sources to double-check where the information comes from.

Book Recommendations

The books I read probably tell more about me than I can imagine. So, I compiled a list of all the books I have read using Goodreads and created a separate list of some of my favorite books. Then I have asked Deep Research to:

• Tell me what beliefs and ideas I might hold about the world, society, and human nature based on the books I read.
• Synthesize the topics that I am gravitated towards and make recommendations for new books that I might like, along with a short description of why I think I might like them.
• Based on my reading list, what blind spots might I have, and what suggestions can it make to challenge my current worldviews?

Results

With this query, I noticed one of OpenAI Deep Research's shortcomings. Even though I provided a list of books I had read, it still recommended books on that read-list. However, Manus was much better at following instructions and making recommendations that were completely new to me. I had this issue with my next deep research on travelling as well.

Travel Recommendations

I love travelling. I have been to more than 20 countries so far. I have used Deep Research for self-exploration based on my travel preferences and to discover new places. I have provided it with a list of cities I have been to, a short list of places I have loved and why I loved them, and a list of places I would like to avoid. I have also given a short list of activities I enjoyed while travelling. Then I have asked Deep Research to:

• Tell me what kind of a traveller I am.
• Recommend places to go that I haven’t visited before in different categories of travel options (like cultural, seasonal, etc.).
• Recommend unique activities based on what I have liked before.
• Recommend specific places to go given the dates of vacations that I have so far for this year.

Results

OpenAI again did a poor job of following instructions and avoiding recommendations that were already marked as visited in the prompt. Manus did a much better job in coming up with vacation options and activity suggestions. I would definitely be using these services for further travel planning.

Health Research

If you lived long enough, you may have a chronic health issue that you are dealing with. I have been dealing with a neurogenic cough for over a year and it took several doctor visits and reading a book on chronic cough to understand that was my problem. OpenAI generated Deep Research on this condition was top-notch and could have been helpful several months ago as I struggled to understand the root cause of my persistent throat discomfort.

Other Research Topics

I had other research topics based on my side project, Heuristi.ca, like competitive research, business recommendations, grants I could apply for, etc. However, the topics I have listed above interest me as they lead to self-discovery and improved quality of life. I would love to learn more research topics like these that I can use for further self-discovery. Feel free to send me an email if you have some suggestions. I have tried asking AI but haven’t yet come across many ideas that appeal to me.

Deep Research Tool Comparisons

Here is my current impressions about the Deep Research offerings by different providers.

Unfortunately, Perplexity and Grok generate output that is neither comprehensive nor great to read. I like how OpenAI asks follow-up questions when you enter your prompt to help clarify and narrow down the research process. OpenAI also generates very readable reports that are a joy to read. I liked the output from Manus as well. It can be slightly less readable compared to OpenAI; however, it feels a lot more comprehensive. Additionally, Manus does a better job of following instructions in the prompt.

One idea is that you might want to refine your research prompt by using something like O1 to make sure it is comprehensive enough. This way you could get the most out of your requested report.

My recommendation would be to definitely give Deep Research from Manus and OpenAI a try to see how it can help you with your life.

References

OpenAI Deep Research: https://openai.com/index/introducing-deep-research/

Manus AI: https://manus.im/