Artificial intelligence is a great tool for research, as long as it is used correctly.
Before I had her, my typical process began with a brief . It consisted of understanding the objective of the research: What do I want to learn? What am I going to do with the results?
From there, I defined my research questions, since the art of good research lies largely in knowing how to formulate the right questions.
With clear objectives and questions, I searched for relevant sources of information: quality articles, books, YouTube talks, and reliable websites. I spent hours reading, watching videos, and taking notes to extract valuable snippets that would answer my questions. Finally, I compiled a report with the most relevant information, adding my own analysis, reflections, and conclusions.
The mistake of skipping the process
When I started using AI, I made the mistake of trying to skip the whole process. I would turn the brief directly into a prompt , expecting it to give me everything ready-made, but I ran into several problems:
- Hallucinations: AI invents data or facts.
- Lack of filters: Although it cites sources, it doesn’t always allow you to choose which ones to trust and which ones to discard.
- Lack of critical thinking: AI does not question ideas or identify incorrect assumptions. It tends to be compliant, saying what one wants to hear instead of challenging initial assumptions.
- Loss of visual context: Much is lost when everything is converted to text. There are details, such as the aesthetics of an element or the nonverbal language of a presenter, that must be seen to truly understand.
- Flat results: Reading only text can be boring. A report with examples, images, and comparative graphs is not the same as a response generated in natural language.
AI as an accelerator, not a replacement
Ideally, the process should be maintained, but AI should be used to accelerate it. The first major bottleneck is finding sources . Doing this manually—reviewing results one by one, downloading articles or books—is too time-consuming on mechanical tasks that don’t add value. This is where AI shines, quickly finding relevant, recent, and reliable sources.
The second bottleneck is processing the information to identify the key fragments. Reading an entire book is valuable, but in a quick investigation it’s not always efficient.
For this, I recommend NotebookLM(from Google).It’s afreemium tool that lets you organize your sources in a “notebook” and ask specific questions or generate audio summaries, mind maps, and presentations. My advice is: follow the entire process. Find the sources, validate them one by one, and build your notebook with only the best ones. Use AI to ask questions, but always verify the answers in the original source (the tool cites them for you to make your work easier). In the end, create your own deliverable that goes beyond just text; include images and original clips to create something that not only informs, but also inspires.
Blaster In-Sight
If you’re looking for something more advanced, In- Sight is our service that does precisely all of the above, but powered by a team of expert researchers and an AI designed specifically for this purpose.
Unlike NotebookLM , our software uses multiple agents and models ( LLMs ) trained with a methodology proven over years. With each client:
- We formulate a customized brief and define the key questions.
- We identify and validate sources one by one.
- We perform data mining and analyze the results.
- We deliver a strategic and professional report in less than 3 weeks.
If you’re interested in taking your research to the next level, let’s talk!
