How to do 10 hours of research in 20 minutes with Ai (Gemini + NotebookLM)

Hey everyone. On today’s show, I’m going to blow your mind, because I’m going to help you become one of the top 5% of AI learners and practitioners in the world — and I’m going to do that by showing you how to stack AI tools together to create super productive workflows.

Today we’re going to break down how you can use Gemini Deep Research with NotebookLM and its sick new interactive voice feature. Using this workflow, I truly believe anybody can be an expert on a topic in minutes instead of days. They can be world-class in weeks instead of years. This is the opportunity to accelerate your life, accelerate your career, your business. Let’s get into this show.

Okay, so this is what I was doing a lot of the holiday break — thinking about how you put multiple AI tools together to do really interesting work. And I wanted to share today how I’ve been using a couple of Google’s AI tools.

One of the tools that I really love is Google Gemini and its new Deep Research feature. I’m a very curious person — I want to get into the details of a topic — and Deep Research has been an amazing way to do that.

So if you go to Google Gemini, what I’m trying to do is: I wanted Google Gemini to do a research project on AI agents — the use of AI agents, the marketing application of AI agents. You can kind of see my prompt there, which is:

“I want you to do a deep research project for me on artificial intelligence agents. I want to know what they’re capable of automating today, what they’re not capable of automating today, when should I use an agent versus a traditional web application, and how should I think about how AI agents should be used in marketing?”

So if you’re watching this podcast — which is all about marketing and artificial intelligence — these are the kinds of questions you’re going to have, and these are the kinds of questions I have.

What I would have done before this is I would’ve gone and done a bunch of manual research and culled it and copied it together — and Gemini is doing that for me.

So what you’ll see right now is that Gemini is working, and it has researched over 32 websites on this topic. You can see all of them there. It has everything attributed, and now it’s in the midst of creating my report. It’s going to take what I want to know, it went out and did all the work.

This is a great example of AI. I probably would’ve only gone to 10 or 15 of those 32 websites. I wouldn’t have found all 32. We’re also going to assume that Google’s AI is as good or better at doing the search queries and finding the most relevant links and articles as I am.

And here it goes — it generates a report for me. That took about five minutes. Before, that would’ve taken me five hours — at least a couple hours — to go through and do everything.

What you’ll see here is that it has put together a very detailed and very robust report with all these topics and sources. And what I love about it is that I can just go and open it in Google Docs and create a shareable document to share with anybody, which is very cool.

So now I have this document here. I could go read through it. I could do everything I want with this. And that would be cool — but I don’t think it’s actually as helpful as we’d want.

So I want to do something different. I want to go to a different one of Google’s tools, which is Google’s NotebookLM tool.

Hey everyone. I’m so excited to share something with you that’s an absolute game changer for marketers like us. We’ve put together the real playbook top marketers are using to turn AI into their secret weapon, and this guide skips all the fluff. It’s packed with strategies that are actually making an impact right now. We’re talking proven tactics straight from the pros at HubSpot, a16z, and Asana. Whether you’re looking to scale your marketing operations or future-proof your career, these strategies are exactly what you need. Grab your copy at the link in the description. Now let’s get back to the show.

So I go over to notebooklm.google.com, and this is what it looks like when you first come in. It’s blank. It’s empty. And the goal is to let you create a notebook of lots of materials that you’re trying to learn about or work with.

What’s cool is: let’s say I want to upload or add sources to my notebook. The first thing I can do is add a Google Doc. So I want to add this AI agent capabilities research — the research that Google Gemini just did for me. So all that deep research we were just talking about, that’s now in NotebookLM. That’s pretty cool.

But I’ve got some other sources I want to add. I’m a big fan of what our friend of the pod, HubSpot co-founder Dharmesh Shah, is doing over at agent.ai. He also has a newsletter called simple.ai — go check it out at simple.ai to subscribe — and he posts all of his newsletters online.

So I can take the links to these newsletters and add them into my NotebookLM. I’m just going in and adding those. That’s pretty cool. So I’ve got a few relevant newsletter articles posted there. And if you didn’t have the links, you could just paste the raw text into NotebookLM.

What I also want to do is: I said I’m a big fan of what Dharmesh is doing at agent.ai, and he was on My First Million where he talked a lot about AI. So I’m also going to take that episode he did on My First Million and add it as a source over here, because NotebookLM allows you to easily add YouTube videos as a source.

And what’s really cool about NotebookLM is: I can have all these sources, or I can interact with just one of these sources.

Okay, so the other thing that is new is NotebookLM has come up with a brand new feature where you can interact with the “podcast.” It can take all of these sources and generate an audio overview.

So I’m going to generate an audio overview. And while that’s generating, I’m going to do a prompt to get a summary.

“I would like a detailed summary from all sources that outlines the five best uses of AI agents today, five examples of how to use them in your marketing, and an executive summary.”

Okay, that’s my prompt. I’m going to submit that prompt. It’s looking at the deep research I did. It’s looking at Dharmesh’s YouTube video. It’s looking at those newsletter examples that I added.

And here it goes:

The five best uses of AI agents today: automated complex workflows. And what’s really cool is it gives you sources and citations. One of the things I think Google has done a really good job of is connecting the AI result back to the source material.

It’s really good at data collection and analysis — unsurprising. It’s great at personalizing your marketing and customer engagement — totally agree, that’s the number one use case we’ve seen at HubSpot. It’s great at content creation. And agents are great at acting as digital proxies.

Agents can understand user preferences and handle basic tasks such as planning travel, evaluating software, managing schedules.

Where you don’t want agents: they lack common sense and real-world knowledge. They don’t have extensive memory. A lot of them don’t have fully up-to-date information. They don’t have the creativity and emotional nuance and emotional intelligence of humans.

So it’s pretty clear what’s good and what’s bad.

And what’s awesome is: from a massive amount of deep research that Gemini did and saved me a bunch of time — plus additional resources that I added because I thought they were really good — I got a very manageable readout on what to do.

Okay, so now what we have is: after a couple minutes of work, it has generated a 23-minute audio overview where I can listen to a 23-minute conversation.

Let me give you a read of what it sounds like.

“Welcome to this deep dive all about AI agents.”

Oh, cool.

“We’re going to be looking at a bunch of articles today, and even a YouTube interview we found…”

“All right, CTO of HubSpot…”

Oh yeah — it has the context to know who Dharmesh is.

“Oh wow. Okay, he’s calling AI agents the next big thing…”

So this is like having a personal podcast. When NotebookLM first launched, this was the feature where we were like, “Wow, this is incredible.” I can get an amazing summary in podcast and audio form so I can consume it on the go.

And then it got even cooler because you could go to Deep Research in Gemini, get rich insights, bring that into NotebookLM, and get it in audio form.

Now what I want to show you today is their brand new interactive mode, which is still in beta, but it allows you to interact with the show. I can join the conversation with the two hosts.

Let’s see how this goes.

“Hey there. Want to share your thoughts on this?”

Yeah, I was hoping you could explain to me how I should be using AI agents in my marketing strategy, based on what you two know.

“Oh, that’s a great question, and something we can definitely dive into for you.”

“Yeah, we were just about to get into the nitty gritty of what AI agents are and how they’re totally different from things like chat bots. Then we can definitely talk about how you might be able to use them in your marketing.”

Yeah, it seems like Dharmesh — the guy we’re talking about — is really excited about AI agents.

“Yeah, he’s the founder of agent.ai.”

“Oh wow. Okay. He’s calling AI agents the next big thing in tech — big statement — and even compared his new company to like the LinkedIn for AI agents.”

“Well, that’s a bold statement for sure. It definitely piques my interest.”

“Yeah, me too. So before we get too far, let’s break down exactly what an AI agent is, okay? Because I think when a lot of people hear the words AI agent, they just think like chat bots.”

“Yeah, right. But this seems like it’s something totally different, right?”

“Yeah, you’re right. It’s way more than just a chat bot. Think of it more like a digital teammate that can handle complex tasks — things that require multiple steps.”

Got it.

When you’re talking about AI agents handling complex tasks, can you tell me more — what are some examples of complex tasks AI agents could handle?

“That’s a great question, and it’s helpful to get into some specifics. So it’s not just about simple one-off tasks like what a chat bot can do. Instead, think about things that require a sequence of steps — like planning a whole business trip. An AI agent could handle booking flights, finding hotels, and making restaurant reservations.”

Wow.

“Or in marketing, they could manage an entire social media campaign. It could start with generating content, then scheduling posts, and even analyzing engagement.”

I’m really interested in that trip-planning use case. What specific tool or tools could I use to actually go do that now?

“That’s a super practical question, and it’s great you want to try this out. While there isn’t one single perfect tool that does everything end-to-end yet, there are some platforms you could start experimenting with today. For booking flights and hotels, some people are using tools that have integrated AI like Kayak or Expedia. These tools can help you find the best deals, and some even offer personalized suggestions based on your past trips.”

That makes sense.

“Then for things like restaurant reservations, you could try something like OpenTable or Resy, which also sometimes have AI features.”

Okay — so no single AI agent to handle the entire process quite yet?

“Not quite yet, but the idea is that these tools are already starting to become more interconnected. We’re seeing the building blocks right now.”

Exactly.

“It’s like we’re seeing the early versions of those multi-agent networks that Dharmesh talked about. And while those specific trip-planning tools are not fully agent-based yet, they give you a good idea of how an AI agent could work.”

Yeah. It’s cool to think these tools are going to keep evolving.

“Yeah. And it’s also important to note that these agents will improve as they work with you and learn your preferences over time. And that’s where this idea of hybrid teams that Dharmesh talks about comes in. He thinks that we’re going to soon be working in teams where you have humans and AI agents working together, and each plays to their strengths.”

I really like this conversation about hybrid teams. What could hybrid teams look like?

“That’s great feedback. We’re glad you’re finding this interesting. Let’s dive deeper into this idea of hybrid teams. We were talking about AI agents taking over repetitive tasks — like scheduling social media posts — and that frees up human employees.”

Give me the five most repetitive tasks that AI agents could handle today.

“Absolutely. First would be scheduling social media posts. AI agents can handle posting across different platforms. Second would be data entry and report generation — agents can pull data and create basic reports automatically. Third is sending out personalized email sequences — they can nurture leads automatically and tailor messages to different users. Fourth, agents can handle basic customer service inquiries, like answering FAQs, freeing up human support staff for more complex issues. And lastly, they could do A/B testing of ad copy and images — trying variations and analyzing results.”

Could you tell me what tools I could use for those five examples?

“Okay, that’s a very practical question. For scheduling social media posts, people use tools like Hootsuite or Buffer. For data entry and reporting, you could check out tools like Tableau or Google Data Studio. For personalized email sequences, Mailchimp and ActiveCampaign are great options.”

What about HubSpot for email marketing? I’ve heard they do a good job.

“That’s a great point. HubSpot does have its own marketing automation tools. You could definitely use HubSpot for personalized email sequences, and it can handle a lot of other marketing tasks too.”

I think what’s interesting here is how they built this feature. You are a listener and you’re asking contextual questions in the moment. You’re jumping in.

If you noticed, the first thing I tried to do was use it like a voice chatbot and ask what I wanted right off the top. And it was basically like, “Nope, nope, nope — we’re going to play the script. And when it comes to the part we’re talking about where it makes sense, we can address your question.” That’s where it really shines.

And if you’re like me, you’re often watching a show on YouTube or listening to a podcast, and you have a bunch of questions. You’re like, “That’s cool, but I need the next level of detail — what about this?” This experience lets you do that.

I suspect not too far away you’ll be able to watch your favorite podcast, and through AI and AI avatars, the hosts will have avatars and interact with you and go deeper on your individual questions, which would be amazing.

But right now, I think Google has built the best stack of products for research on the planet. If you’re trying to do great research, I think the combination of Google Gemini with Deep Research and NotebookLM is unmatched.

You do deep research, put that into NotebookLM, add any other sources, query that information, and learn in the way that works best for you — including this new voice beta feature where you can interact with the generated audio overview.

What’s really cool is you can still listen to the podcast while you’re working. And if you have questions, you can just chime in.

Where this really shines is going from novice to high-level expertise on a subject exceptionally fast. That’s what we’re talking about today — going from knowing very little about something to being a top 20% expert in days or weeks, not months or years. That is transformational in terms of your ability to find, access, and process information.

These tools have been awesome at it.

I highly encourage you to go play with the new NotebookLM interactive audio feature. I find it to be one of the better voice AI features I’ve used so far. I think they’ve done a really good job.

I would go use Gemini Deep Research and NotebookLM together if you’re tackling any projects where you’re trying to learn and get up to speed. This will help you get started so much faster.

Thanks everybody for tuning in to today’s episode. See you next time on Marketing Against the Grain.