
Loading summary
A
On this episode, I'm going to show you how to create two AI assistants to make you the most productive person in your company. I'm going to show you how to build a executive assistant and a project assistant. And better yet, I'm going to show you how to do that across Gemini Gems, ChatGPT and Claude Projects. And I'm going to compare and contrast which are better. Which one should you actually use for what? All of that and more on today's show. I'm Kieran Flanagan, the co host of Marketing against the Green. Let's go build some AI assistance.
B
Here's a quick message from HubSpot. This isn't your typical marketing software ad because HubSpot isn't typical. It's marketing made easy. Turn one piece of content into assets for every channel, convert leads in no time and get a crystal clear view of your campaign performance. HubSpot can do all of that and get you results fast. Like double your leads in 12 months fast. See, I told you this wasn't a typical Software ad. Visit HubSpot.com marketers to get started for free.
A
So this is a really great video if you just want to get started with the AI. Like I think this is one of the easiest things you're going to be able to replicate for your day to day work. Because it's not my day to day work as a marketer, it's not my day to day work as a special whatever I do. It's really just how we do work. And I'm going to show you how I actually started to use AI to be way more productive. And so let's get into it. The two AI assistants I'm going to build is an executive assistant and a project assistant. So people who follow along my videos may have seen me build an executive assistant for Claude. Well, I'm always building upon and iterating upon what I've done in the past and I've made that really much better. And I've also replicated it for the three core platforms, Google, Gemini, OpenAI, ChatGPT, and Claude. And I'm going to show you a comparison because I think it's pretty interesting in comparison, like which ones are better for your day to day work as a knowledge worker. The first thing I actually want to start with is something you probably haven't even thought about for AI, but you should. Here's a little sneak peek at something I do that I think you should start doing. And no one else would tell you this tip because it's so mundane. It's actually folder structure. If you are using Google Drive or whatever folder you're using, I'm going to show you how to use your folder structure to build these AI assistants. I am someone who is pretty disorganized in terms of like where I keep things. And because of AI that actually no longer matters, it doesn't really matter where I keep things. So there's two things you have to do when you're building full structures and the way I would do this. So for me in HubSpot, I have three large pods that I oversee. I have our go to market in the lower end of our business. We have our overall demand just how HubSpot's go to market works and then we have all of our AI work. And so what I do is I have a folder structure that basically just has two different folders of stuff. I have one where I put every single meeting transcript, record everything, put the meeting transcripts in there, and then all of my strategic docs. And what I'm going to show you is the reason I do that is I build an AI executive assistant for the meeting transcripts and I build an AI project assistant for all of the strategic docs. So that's kind of tip number one. I always start to think about how your folder structure allows you to build these kind of AI assistants. And I think to follow along my video, this is my recommended structure. One, whatever your project, break it into meeting transcripts, which is all of the meetings you're having on Zoom or whatever else, and you're keeping the transcripts and then the strategic docs. Now, at some point I will do a follow on to this video where this is not just Zoom meeting transcripts. It's going to include emails and it's going to include slacks. But for the time being, I think that is where I would start. The other cool thing is, you know, a lot of people, what you'll do is go to Zoom. I'm going to use Zoom because it's the most common. But whatever you're using, you go, you record the meeting, you get the transcript and you have to put it in the folder. Or what I'm doing is I have set up Zaps. So every time the meeting ends and the transcript is available, it just zaps it into the right folder. So that's number one. Okay, let's go into Build this in Gemini because we haven't actually built Gems. I focused a lot on Claude. I'm still going to do Claude in this video, but we haven't actually talked much about Gemini Gems Gems are basically Google's equivalent to custom GPT. So when you go into Gemini you'll see the gem manager here. So basically you can see all your gems. So they have pre made for Google. Very similar to ChatGPT store. I built two, I built our project assistant and I built a scaled selling assistant. So these are for one of my pods and so I go into the project assistant and so one of the things I can do now is show me the last summary of our meeting, including key follow up points, taking a little bit of time to load, analyze and response. And it will go through and it will show me my last meeting and tells me everything I need to know. And so let me go through how we've actually built this. We go to Gem Manager, we're going to edit Executive assistant. And so what you do is you build an executive assistant and this is the prompt here. You basically tell it to do meeting analysis. So anytime you give it a meeting transcript, it analyze the meeting transcript. It identifies key decisions and action items, it tracks deadlines and ownership, it flags high priority items and then I give it the exact output format. Now if you want these prompts, you sign up to my newsletter. I'm going to give you this prompt and the project assistant prompt. You can find my newsletter sign up in my LinkedIn. And so we'll go back to Gems. And so I put the prompt in and then the next thing you can do is you can basically upload everything from your Google Drive. And so if you have a folder structure. So again, what did I say? All my folder structure looks pretty the same. And, and then you can go to meeting transcripts and grab all the meeting transcripts. Now word to all of the people building these AI assistants, Google, ChatGPT and Claude all do a bad job of this in that you have to kind of go in and upload the files or you connect to your Google Drives and you go add them. I have to go in here and add the new meeting transcript in here to get all of the follow up actions. What I want to do in the future is be able to just zap it straight into this context window. So every time I come back into my assistant it has the latest meeting. So that's it. So now you have an agent to be able to tell you anything that's happened in any of your meetings. It has access to all of your meetings. Just remember to go in here, add them from your Google Drive. I've already done that. You can go in, add any meeting transcript, add the latest One. So that's the only piece of work you have to do is like go, remember to do the meeting transcript. Now the next one is super cool. The project assistant. So this is a strategic project assistant and it does some pretty cool things. So it basically says you are an AI project assistant with expertise in project management, risk analysis and organizational dynamics. Now remember, when you get this prompt, if you follow along in my newsletter and you really want to kind of replicate this for yourself, you can actually play around with this prompt, right? You should just not copy and paste for your needs. You should use it as a foundation and then build upon that foundation and tweak to your needs. Again, the meeting exec1 has access to all of your meeting transcripts. Remember the folder structure. Now your project assistant has access to all of your strategic docs. And what it does is basically it presents a menu of available services. So anytime you have a new strategic doc, you put it into that folder. And what it will do is it will analyze all the documentation and basically give you a summary of their project, including some really cool things, right? Team structure, responsibilities, any timelines, all of the budget, risk assessment, technical specifications, stakeholder communication logs, previous status reports and minute meetings. This one here I'm trying to think through. Someone may argue why don't you just combine these things in two. It's actually better to keep them separate. They work much better if you have one very focused on project management, one very focused on beta notes. And so this is doing a little bit of like executive assistant where it is grabbing previous minute meetings. But I'm probably going to take this out and keep it very centric to project management work now and then it provides actually a menu from you to pick from. So you can basically ask it to do further project status analysis. And I'll show you what that looks like. You can do a risk assessment which is really good. It like pinpoints any potential risks. It does a really good job. Well, some better than other. I'll actually show you the three perform very differently when I give it this prompt. Tactic generation. So we'll actually pull out the core components of your project and recommend different tactics to make that component a success. Strategic recommendations. If you're having to put something together for your executive team, you can choose us. It does a really good job and then any kind of document generation. So I can basically create a monthly summary, a quarterly summary, a quick summary for the executive team for our CEO really easily because it has all of the documentation. The minute meetings compilation. Again, I'll take this one out because I've actually since progressed to split these things out. I had it all in one and then action item tracking, decision log maintenance. Like it gives you the kind of history of how a decision was made, which I think is really important. And then progress summaries, which is really important. Now I can ask it for specific things. It actually is very good. It gives me some really interesting things I hadn't thought about. But I'll show you what I mean. Like let's create a one page summary for my executive team detailing key accomplishments and challenges. All right. Gives key accomplishments, give some key challenges. Looking back over all of the presentations and docs that we've done. Overall status like pretty good length. So it actually does a really good job of being your project assistant. Now for any project I have a meeting assistant for that folder and then I have a project assistant for the strategic docs folder. And again, just going back to that project assistant, really what we want to do is have it summarize all of those strategic docs anytime you want. Now when you're summarizing the strategic docs, you can ask it for specific parts of the docs, like show me what is different from this month to last month in terms of the progress we've made. Show me that summary. Then you can get into the actual what are the options you want? This is the one that actually is really good because I was trying this where it actually shows you key components of a project and will show you tactics. But then it will do anything for you. Like it will do all document generation, it will do strategic recommendations, it will do all of that stuff. Now that's the way I'm sending it up. The reason I've kind of glossed over some of the output is because this is actually my real assistant, right? So it actually has access to a lot of real documentation. Again, the documentation you have to go in to Gemini. What I really like about Gemini is the context window is really large. So you just have to go in and remember to continue to like add the latest docs. There's no way yet to like automatically add the docs into that context window. Now if you're watching along LinkedIn, this is where I will leave you because the LinkedIn video can't be too long if you are subscribed to my newsletter. If you are a Marketing Against Green Podcast listener, you're going to get the full episode where now I'm going to go on and show you how to do this in Claude, how to do it in OpenAI and one of those is actually even better than Gemini. So Gemini is really good. Very, very good actually. But one of the ones I'm going to show you is surprisingly amazing.
B
Let me tell you about a great podcast. It's called Creators of Brands. It's hosted by Tom Boyd. It's brought to you by the HubSpot Podcast Network. Creators of Brands explores how storytellers are building brands online. From the mindsets to the tactics to the business side. They break down what's working so you can apply that to your own goals. Tom just did a great episode about social media growth called 3K to 45K on Instagram in one year, selling digital products and quitting his job to go full time creator with Gannon Mayer. Listen to creators or brands wherever you get your podcast.
A
All right, so how does this look in ChatGPT and Claude? And so if we go into OpenAI and we create some custom GPTs, I would say that again the getting the documentation is not fabulous. So what I've done is created custom GPTs, I've given some prompts. I can just click and say analyze the transcript for key decisions. I can do what are the high priority items? Can you provide a stormy and follow up, identify the action items with ownership, which is really great. Exact same prompt. There is no difference in the prompt here. The only thing I will say is again the best way to use this is after a meeting again you can add it to your G drive and then I can go in and I can grab my median transcripts. So let's go here, let's grab. You just have to go in and you add your latest meeting transcript and then you add that doc. So again it will be much better at some point if it just has access to a folder where those folders are keeping your documentation so you don't have to go in and just add it. And anytime you want a meeting summary you can just go in and ask for any kind of meeting summary. Okay, so this is the scaled silent executive assistant. So it works very similar way. So It's a custom GPT. So I go into Explore GPTs, I go to my GPTs, I have the exact same two assistants. Assistants, the same prompts. So if I go into the executive assistant I can upload a meeting transcript here again I've got it connected to my G drive and then I can upload this and I can say analyze the meeting transcript for key decisions and action items. So anytime I need anything around a meeting I've got it recorded. It's in the folder I Can go in, I can upload it here. So it does an incredible job of getting you the meeting summary I think better than most meeting tools actually. So I really never need to go and figure out what is happening because I record all meetings. Everything is recorded from like one to ones to team meetings. So I have this like database of unstructured data that I can go and get insights from anytime I want. The one that's much better here in OpenAI ChatGPT than I think Google Gemini is actually the project assistant. Now again, what I really get frustrated on is just the way you have to add the docs. So let's say I wanted to use my assistant here to get some worked on my project and I'm going to add all of these documents. So add all the latest strategic docs. Again, not to continue to repeat myself. Be so much better if it could just auto access a folder so I didn't have to add the docs. But then I'm going to say provide a gap analysis of the project plan. So like based upon our current course, where do you think are our gaps and challenges? Takes a little bit of time and so it will do a pretty great job. And so if you actually do that, it'll give you like an overview, strategy goals, it gives you lots of the strategic gaps, operational gaps, execution gaps, builds you a whole table of potential risk and integration strategies, recommendations for short terms and long terms. It's pretty unbelievable to be honest with you. So you really do have a product assistant that can be strategic with you and actually go back and forth. And so the other one that works really well is when you ask it for like show me tactics around components that you think are the most important to make this thing a successful. So I would say ChatGPT is not much better, but is better in terms of the project assistant if you build that versus what I've seen on Gemini gems. Both very useful and you can prod and poke Gemini to be better. If I ask it for more comprehensive details it will actually do that. But that's how you use a project assistant. You build custom GPTs. And now again, when I'm finished, I'll have three sets of those. A set for each folder, a meeting executive and a project assistant. And then the last one I'll just quickly cover is Claude because I want to talk a little bit about some of the actual drawbacks of Claude for some knowledge work like this. So I would say at the moment like ChatGPT number one, mostly just because of its output in terms of this Project Assistant, it just does better strategic thinking around that for some reason. The two I have here, I'll go into Project Assistant Claude's core drawback here is I have them all connected to my G drive because I use them all at work. But it doesn't have the ability to pick up PowerPoints like Google Slides just has text unless I'm doing something wrong. But it would not find any of the PowerPoints. And then also the context window is just too small. So when you're building them in cloud projects, you go in and you actually add in your project instructions, right? And it's the exact same prompt and then you add your content. And again, I don't know why they've done it this way. If you see the other two, I can just open up Google Drive and go pick the files here. You can't even do that, right? It doesn't allow me to like just open Google Drive to go and actually pick the files. But let's say I wanted to give it some strategic docs, like I can add them in here. Yeah, you have to add them one by one. So I think the context window is smaller and being able to add the documents is not as intuitive. And I cannot actually add in PowerPoints. But it actually gives you pretty great recommendations. I would say the recommendations it gives you for these two prompts specifically for the Project Assistant is actually a little bit better than Gemini or maybe on a par with Gemini with ChatGPT being a little bit better. So that is the video. The main takeaways here is actually not the output because the output is really good. It's how I did it, right? I structured the folders, meeting transcripts, strategic docs, and then I built an assistant for ea, built an assistant for Project Assistant and then I built that menu of options that you can actually use it to get a summary. You can use it to do further things on the project for you. That's a really great way to start to actually integrate these into your day to day work and actually makes a big, big impact. I use them every single day. I kind of like gravitate between ChatGPT and Gemini because Gemini is so deeply hooked into the Google suite of products. It's really useful, Claude. I use it every day. I use it actually to take some of the documentation that the assistants build for me around my projects and I ask Claude to rewrite them because it's a better writer. So I'll take the output from Gemini Gems on the one pager it built for the executive team and Then I'll put it into Claude and I'll say, hey, can you rewrite this for my executive team? This is the audience, this is what I'm trying to do. So I use all three in some sort of combination. But if I was going to end this in terms of a summary, all of them are pretty good to create a first pass at a meeting executive. They all do a really good job because all of the meeting transcripts are text files. Project assistant dependent upon what documentation you use and what you want to put in there. I would say Google Gemini Gems is really good. It has access to all types of documentation. Really large context window, easily hooks into G Drive. Can just add them in there. ChatGPT really good. The output does a slightly better job of like strategic analysis, strategic recommendations and then Claude really struggling because you build projects, you have the context window. Context window a little too small and actually harder to upload the documentation you want. So that is today's video. I hope this is a good way for you to think about how to start to integrate AI into your own work. And until next time, enjoy this.
Marketing Against The Grain: Episode Summary
Title: I Built 2 AI Assistants That Made Me 10x More Productive | Step-by-Step Guide
Host: Kieran Flanagan, Co-Host
Release Date: February 6, 2025
In this episode of Marketing Against The Grain, co-host Kieran Flanagan delves into the creation and implementation of two AI assistants designed to drastically enhance productivity within a company. Kieran outlines a comprehensive approach to building an Executive Assistant and a Project Assistant using leading AI platforms: Google's Gemini Gems, OpenAI's ChatGPT, and Anthropic's Claude. He emphasizes the importance of choosing the right platform based on specific workplace needs.
Kieran Flanagan [00:00]: "I'm going to show you how to create two AI assistants to make you the most productive person in your company."
A foundational tip Kieran shares is the establishment of a robust folder structure. This organizational method serves as the backbone for the AI assistants, facilitating seamless access and management of data.
Kieran Flanagan [03:15]: "I always start to think about how your folder structure allows you to build these kind of AI assistants."
He recommends segregating data into Meeting Transcripts and Strategic Documents, enabling the Executive Assistant to handle meeting-related tasks and the Project Assistant to manage strategic initiatives effectively.
To streamline the process, Kieran employs automation tools like Zaps. These tools automatically transfer meeting transcripts from platforms like Zoom into designated folders, reducing manual effort and ensuring data consistency.
Kieran Flanagan [04:10]: "Every time the meeting ends and the transcript is available, it just zaps it into the right folder."
Kieran explores the functionalities of Gemini Gems, Google's custom AI solution akin to ChatGPT. He details the process of creating an Executive Assistant by configuring prompts that enable the AI to analyze meeting transcripts, identify key decisions, track action items, and prioritize tasks.
Kieran Flanagan [06:00]: "You basically tell it to do meeting analysis. So anytime you give it a meeting transcript, it analyzes the meeting transcript."
The Project Assistant is similarly developed to handle strategic documents, offering summaries, risk assessments, and strategic recommendations. Kieran highlights the significant advantage of Gemini Gems' large context window, which allows the AI to process extensive documentation effectively.
Kieran Flanagan [09:30]: "What I really like about Gemini is the context window is really large."
Kieran compares Gemini Gems with other AI platforms, particularly ChatGPT and Claude. He demonstrates how to create custom GPTs in ChatGPT, replicating the functionalities of the Executive and Project Assistants. ChatGPT excels in strategic analysis and provides comprehensive project assessments, often outperforming Gemini in this aspect.
Kieran Flanagan [11:10]: "In OpenAI ChatGPT... the output does a slightly better job of like strategic analysis, strategic recommendations."
When discussing Claude, Kieran points out its limitations, such as a smaller context window and challenges in integrating non-text formats like PowerPoints. Despite these drawbacks, Claude remains useful for specific tasks, such as rewriting documents for better clarity and presentation.
Kieran Flanagan [10:30]: "Claude's core drawback here is... the context window is just too small."
Kieran provides a comparative analysis of the three platforms:
Gemini Gems: Best integrated with Google Suite, boasts a large context window, and handles diverse documentation seamlessly. However, it requires manual addition of documents into the context window.
ChatGPT: Superior in strategic thinking and recommendations, making it ideal for in-depth project analysis. Like Gemini, it also necessitates manual document uploads but offers more nuanced output.
Claude: While useful for specific tasks like document rewriting, it struggles with larger projects due to its limited context window and less intuitive document integration.
Kieran Flanagan [10:45]: "ChatGPT really good. The output does a slightly better job of like strategic analysis, strategic recommendations and then Claude really struggling..."
Kieran shares practical insights into integrating these AI assistants into daily workflows. By maintaining a well-organized folder structure and leveraging automation tools, users can ensure their AI assistants remain up-to-date and efficient. He advocates for using a combination of AI platforms to capitalize on each one's strengths, such as using Gemini for data management and ChatGPT for strategic analysis.
Kieran Flanagan [10:55]: "So I'll take the output from Gemini Gems on the one pager it built for the executive team and then I'll put it into Claude and I'll say, hey, can you rewrite this for my executive team?"
Kieran concludes by reiterating the transformative potential of AI assistants in enhancing productivity. The key takeaways include:
Kieran Flanagan [12:30]: "All of them are pretty good to create a first pass at a meeting executive. They all do a really good job..."
By thoughtfully implementing these strategies, professionals can harness AI to streamline workflows, enhance project management, and drive organizational success.
Notable Quotes:
Kieran Flanagan [00:00]: "I'm going to show you how to create two AI assistants to make you the most productive person in your company."
Kieran Flanagan [06:00]: "You basically tell it to do meeting analysis."
Kieran Flanagan [09:30]: "What I really like about Gemini is the context window is really large."
Kieran Flanagan [11:10]: "In OpenAI ChatGPT... the output does a slightly better job of like strategic analysis, strategic recommendations."
Kieran Flanagan [10:55]: "So I'll take the output from Gemini Gems... and I'll put it into Claude..."
Kieran Flanagan [12:30]: "All of them are pretty good to create a first pass at a meeting executive."
This episode provides a valuable roadmap for professionals seeking to integrate AI into their workflows, offering practical steps and comparative insights into the leading AI platforms. By following Kieran's guide, listeners can unlock significant productivity gains and streamline their organizational processes.