This is, by far, the best article on the topic of GTM i read YTD! thanks a lot. I use Deep Research a lot, e.g. for market research and positioning. But still I got a lot of new ideas from your article!!! Thanks for that!
Glad you found the examples helpful! I have a hard time trusting posts on AI that don’t “show their work”, so I always try to show real examples and include the actual prompts and outputs.
Let me know how it goes! The 5 examples are just scratching the surface, but the prompt template should be flexible enough to do tons of other things as well.
Thanks so much for sharing. Torsten, I was trying to use agent mode to study one of my competitors' LinkedIn ads library, but it says it's blocked by the robots.txt file on LinkedIn. Are you able to get past that?
Because I feel like this rule was implemented a year ago, so I was curious how you got past that.
Interesting, I’ll look into this later. I watched my agent do this multiple times live, and he (it?) navigated the library without problems. I’ll get back to you
This is really a deep research on the deep research tool. I have been using deep research tool for quite sometime but the things you have outlined are something I think are really helpful and will definitely use in my deep research every time. Thanks for such an insightful article.
This is great, thanks! I'm following a very similar approach to use Deep Research for non-fiction literary writing, with some minor changes. I'm sharing the process in my blog.
I start by asking AI to explore a topic from one - or few - inicial references that may have triggered my interest on the topic. I ask AI for a summary of these initial papers and from that to explore the topic's ecosystem in order to find and update relevant sources based on quality criteria. As a part of this task I specifically ask AI to identify lines of thoughts that may have alternative positions on the topic to make sure I get the whole picture/debate and avoid bias to the postulates of the initial references. As a result, I get a research brief.
I manually annotate this initial research brief, to specifically identify statements and arguments that I may require to investigate in more detail, and use this annotated research brief as context for the deep research plan.
Then I ask for a research plan as you suggest, based on source quality and proper representation of relevant disagreeing voices plus my annotated tasks. I use this plan for the final deep research job.
While I'm quite satisfied with the results (some examples in my posts), my main issue is to keep references and citations coherent along the process. Claude and GEMINI struggle very much with this. Any thoughts? Thanks!!!
All the steps, caveats, massaging techniques etc required to get this working is exhausting. Are you sure you're saving much time or are you too impressed with your own cleverness and not realizing that it's just as time consuming? You might as well open google, look for some good sources, plan your own research and then write it up, with the added bonus that you won't be training your brain to be lazy and normalizing instant results in your workplace to the point where your boss expects 10x the work in less time and for less pay.
One of most boring tasks of GTM (according to me alone) is provisioning user to 5 different tools and removing access from same five different tools. As a first step used SSO which is amazing in saving time as well as maintaining single password. Any suggestions to go step further to automating (not an expert on SCIM)
This is, by far, the best article on the topic of GTM i read YTD! thanks a lot. I use Deep Research a lot, e.g. for market research and positioning. But still I got a lot of new ideas from your article!!! Thanks for that!
Wow, thanks so much! Would love to hear some of the detailed use cases or prompts you’ve been experimenting with.
A very detailed and insightful article on AI & GTM research. Thanks for sharing .
love it especially the examples
Glad you found the examples helpful! I have a hard time trusting posts on AI that don’t “show their work”, so I always try to show real examples and include the actual prompts and outputs.
This is super useful! Really looking forward to applying this!
Let me know how it goes! The 5 examples are just scratching the surface, but the prompt template should be flexible enough to do tons of other things as well.
Asking for any clarifying questions at the end of the prompt has also helped me provide more context and improve the quality of output
Thanks so much for sharing. Torsten, I was trying to use agent mode to study one of my competitors' LinkedIn ads library, but it says it's blocked by the robots.txt file on LinkedIn. Are you able to get past that?
Because I feel like this rule was implemented a year ago, so I was curious how you got past that.
Interesting, I’ll look into this later. I watched my agent do this multiple times live, and he (it?) navigated the library without problems. I’ll get back to you
This is an excellent breakdown of Deep Research for GTM!
Wow. Thanks for sharing!!
This is really a deep research on the deep research tool. I have been using deep research tool for quite sometime but the things you have outlined are something I think are really helpful and will definitely use in my deep research every time. Thanks for such an insightful article.
This is amazing Torsten and Kyle. Thanks for sharing. Added to my swipe files.
This is great, thanks! I'm following a very similar approach to use Deep Research for non-fiction literary writing, with some minor changes. I'm sharing the process in my blog.
I start by asking AI to explore a topic from one - or few - inicial references that may have triggered my interest on the topic. I ask AI for a summary of these initial papers and from that to explore the topic's ecosystem in order to find and update relevant sources based on quality criteria. As a part of this task I specifically ask AI to identify lines of thoughts that may have alternative positions on the topic to make sure I get the whole picture/debate and avoid bias to the postulates of the initial references. As a result, I get a research brief.
I manually annotate this initial research brief, to specifically identify statements and arguments that I may require to investigate in more detail, and use this annotated research brief as context for the deep research plan.
Then I ask for a research plan as you suggest, based on source quality and proper representation of relevant disagreeing voices plus my annotated tasks. I use this plan for the final deep research job.
While I'm quite satisfied with the results (some examples in my posts), my main issue is to keep references and citations coherent along the process. Claude and GEMINI struggle very much with this. Any thoughts? Thanks!!!
All the steps, caveats, massaging techniques etc required to get this working is exhausting. Are you sure you're saving much time or are you too impressed with your own cleverness and not realizing that it's just as time consuming? You might as well open google, look for some good sources, plan your own research and then write it up, with the added bonus that you won't be training your brain to be lazy and normalizing instant results in your workplace to the point where your boss expects 10x the work in less time and for less pay.
So what you’re really saying is you figured out how to google properly. 🫡
I think an agent reviewing an ad library and summarizing its findings, or writing a detailed step-by-step guide, goes way beyond googling 🫡
Thanks for article on AI for GTM.
One of most boring tasks of GTM (according to me alone) is provisioning user to 5 different tools and removing access from same five different tools. As a first step used SSO which is amazing in saving time as well as maintaining single password. Any suggestions to go step further to automating (not an expert on SCIM)