The 5 AI Tools You Need After ChatGPT (that do real work)

Opus was the largest smartest model at the time. OpenAI is the most advanced and the most widely used AI platform in the world. Now — Gemini is our largest and most capable model. — GR for is smarter than almost all graduate students in all disciplines simultaneously. — Introducing notion 3. 0. It's the world's most advanced knowledge work agent. — We're going to do something near the frontier. I think better than any current open source model out there. This represents the next step on our journey towards the ultimate goal of infinite context. — Chat GPT Gemini Opus Chat GPT Gemini GPT Gemini. — If you're feeling overwhelmed by AI tools, you're not alone. Figuring out which tool works best for which task is impossible without lots of trial and error. So, in this video, I distill three years of learnings down to just one thing, what each tool does better than the rest. In part one, we covered everyday and specialist AI. So today we'll go over productivity and creative AI. Let's get started. Kicking things

off with productivity AI. First up, Google Workspace. Yes, in part one of the series I covered how Gemini is the only big three chatbot that can natively process text, images, audio, and video together. Here we're focusing on Gemini's deep integration within Google Workspace. In a nutshell, the superpower isn't the model itself. It's the native integration. Put simply, Gemini is the only AI that can search and synthesize across your entire Google workspace from emails to docs to calendar invites in a single query. Although you can technically connect Chachet and Claw to Google platforms, those are third-party connections that connect from the outside, meaning they can be flaky and they can't access certain file types like Google Sheets. Gemini, surprise, doesn't need to connect because it's already natively integrated. So, the apps talk to each other through Gemini seamlessly. Let's go through a real world example. At Google, by the time I was done with a marketing campaign, I would have over 50 meeting transcripts, each one barred inside a separate calendar invite, plus hundreds of pages of notes in shared docs and maybe 200 plus email threads across different teams. Previously, I'd have to spend a week or two digging through all that manually to summarize learnings, draft a recap email, and prepare a debrief presentation. Now, I can just use the at@ workspace extension and ask Gemini, for example, to identify and locate all relevant documents and emails related to this specific project. analyze and deconstruct all the information to understand the campaign's purpose, goal, and results and draft a detailed document I can use to prepare a debrief for the team. It pulls from Gmail, Drive, and Calendar in one single query. And that level of synthesis is only possible because the integration is native. So, as a rule of thumb, if your work lives in Google Workspace and you need to pull from multiple sources at once, ask Gemini first. Side note, I'm developing a full course called the AI systems academy. It's not a tool specific course. It's a system course that happens to use AI, which means what you build in five to six hours will work for years and not just until the next tool update. Link to the wait list down

below. Next up, Notion AI. Notion AI superpower is its agent capabilities, meaning it takes action inside your workspace, but simply instead of just answering questions, it can actually do stuff. Just for comparison, yes, Gemini is able to draft something within an empty Google doc, but and this is very important, but Gemini cannot create and populate documents or spreadsheets from scratch, let alone reorganize them. Now, taking a look at notion AI, can I draft something from within an empty page? Yep, just like Google Docs. Okay, level one, basic capabilities, check. Level two, here I have a database containing one entry. It's a job description page for an operations manager role. And let's say I want to hire a customer success manager. I can open up notion AI and type create a new job opening in this database for a customer success manager based on the operations manager template I referenced that and the notes from the customer success manager role status equals active date posted equals one year from today notion AI picks up the structure the format and even the tone of voice from my existing job description creates a new page and even adds a new tag active within the status property moving to level three notion's relations property is extremely powerful. For example, within my devices and storage area page, if I scroll down, create a new note. Hello YouTube, you will see that this note has automatically been linked to the devices and storage area page. Now, if I wanted to merge this area page with my housing area page, I can simply tell notion AI, hey, I want to merge the devices page with this one. Your task is to move all the related notes, resources, and projects from the devices area page over here. And this saves me so much time. I don't want to waste your time right now because this is not a notion tutorial. But let me know in the comments if you want a full video on how I use Notion AI. And I also leave a link to my full Notion course for beginners down below. Also, let me be very clear. Buying Notion AI does not mean you get access to three models for the price of one. Similar to Perplexity, Notion AI uses fine-tuned versions of Chacht, Gemini, and Claude and optimizes them for notion workspace, meaning they're not as powerful for everyday use. All that said, as a rule of thumb, if you need AI that can build and edit inside your workspace and not just search it, Notion AI is the only option so far that actually takes action. Rounding out

productivity AI, Whisper Flow superpower is extremely accurate voicetoext transcription. This means we naturally give AI richer context than we'd ever bother with typing. Put simply, the transcription from Whisper Flow is not perfect, but it's 95% there. And because it's so reliable, it makes a material impact on how we interact with AI. Jumping right to an example, remember the marketing campaign recap prompt I mentioned in the Google Workspace section? If I just type that prompt out, I'd probably take around 5 to 10 minutes, write two paragraphs, and think, "Okay, this is good enough. I can stop. " With voice prompting, I can talk and basically brain dump for 30 seconds and include all the details I would have otherwise skipped, like specific teams involved, the timeline, and the tone I want. I'm able to easily give more context because the friction of typing is gone. Right? While I am super bullish on Whisper Flow right now, there are two massive caveats. First, the iPhone experience sucks because you need to swipe between Whisper Flow and the app you're using every time to activate it. So, if you're not a power user like me and you have an iPhone, I just can't recommend it right now. Second, I'm not too sure about their long-term mode because it wouldn't take much for OpenAI, Google, or Anthropic to improve their native voice modes to match this level of intelligent auto editing. There are literally thousands of examples of big tech crushing startups by adding a feature into an existing product. All right, moving on to creative AI. First

up, MidJourney superpower is complete and total control. Put simply, MidJourney gives you the most precision over your image output, but that comes with a learning curve, so it's really made for power users. Think of the default camera app on your phone. Most other image tools can be thought of as auto mode, where you basically point and shoot, and 99% of people are happy with the results. Midron is the equivalent of manual mode, where you can dial in the shot by changing settings like aperture, exposure, and styles. And just like how most people have no idea what aperture means, midjourney's syntax means either. Diving right to an example, for this first image, I use natural language to describe what I want. A professional woman giving a keynote speech on stage, modern conference, dramatic lighting, photorealistic. And if we fast forward a little bit, this is what we get. Here's the same prompt, but with syntax. A professional woman giving keynote speech on stage. Blah blah. AR 16 to9 style raw styliz 50 V7 reference this URL no audience faces text and logos and fast forward a little bit we can see just how much of a difference those extra parameters make even when the initial description stayed exactly the same I won't dive into each one of these parameters but the 16 to9 sets the aspect ratio in this case horizontal the S ref one uh locks in a visual style from a reference image and uh low tells Midjourney to exclude specific elements. To be very honest, I pay for Midjourney, but I mostly use it for research, not generation since the platform attracts power users, right? The community gallery is an amazing resource for inspiration. I find what I like, then recreate it with a simpler tool. So, as a rule of thumb, if you want maximum creative control and you're willing to learn the syntax instead of just chatting with a bot, MidJourney is the

industry standard. Next up, we have Google's Nano Banana Pro. And the superpower here is natural language precision editing, meaning it has accurate text rendering and the ability to iterate without starting over. Put simply, if midjourney is Excel, Nano Banana Pro is Google Sheets. It's simpler, it uses natural language, and for most people, it's more than enough. Starting off with a simple example, we'll enable the create images feature in the Gemini app and say something like, "Create a minimalist infographic image that can be dropped into a slide with no further editing based on the following description. " And then I paste a script segment from one of my previous videos. This first output is fine, but there's a lot going on. It looks ugly. So I continue with first remove the box at the very bottom. Second, apply Apple aesthetics and branding colors. Third, optimize for one:1 square dimension. And just like that, I get something that is ready to share. Notice that I didn't have to start from scratch, right? I just continued iterating on the previous image. Use case number two. I drag in an existing photo of a contact lens and use plain English to first transform this into a high-tech smart lens. Notice nothing else in the photo changed. Hence the precision editing superpower. Then I can iterate until it matches what I initially visualized in my head. Pro tip for 4K highresolution image outputs, you're going to have to use Google AI Studio and link a credit card to pay for API usage. Here you can see that I'm using the Nano Banana Pro model. I uploaded a photo myself, set the resolution to 4K, describe the image I want, and I got an honestly mind-blowing thumbnail image that I actually used for one of my videos. For all the AI nerds out there, it's important to know that Nano Banana Pro is one of four image models from Google, and each one has its own best use case. As a rule of thumb, though, the Nano Banana Pro model is best for making precise edits like changing text or colors using plain

English. Moving on, we have OpenAI's GBT image model. And the superpower here is memory. While Nano Banana Pro is great at precise edits on a single image, GBT image excels at maintaining consistency across a sequence of images. Diving right into an example, and don't judge me for this. I use the same prompt in both Gemini and Chacht to generate an anime image of two characters from my favorite games. These styles are clearly very different, but both did a decent job. And note the white strand of hair on the female character and the textile from Chachet's output. It'll be relevant later. Within the same chat threads, I input the next prompt that logically follows from the first. And both Gemini and Chachi generate images that are consistent with the first. So far so good. But look what happens when I send a third prompt that asks for the same characters to appear in a completely different context. Gemini starts to lose that consistency, right? It's hard to tell which one is a female character, while Chachi BT maintains consistency with the white strand of hair and text style. By the fifth prompt, Gemini completely falls apart. This last image was supposed to be the male character protecting the female character from a dragon, but Gemini mixed in elements from the previous prompt, making the image nonsensical. So, what are the implications for the real world? Let's say you're building training materials and you need a mascot that appears across different scenarios. you're going to have an easier time with ChachiBT getting that character to stay visually consistent across all of them. So, as a rule of thumb, if you need multiple related images where characters or visual elements need to stay consistent, Chad PT is the more reliable choice.

Rounding out creative AI with Google Flow, the superpower here is generating images and animating between them without ever leaving the app. Here's how it works. Within Google Flow, we can generate videos and images. And surprise surprise, the native image model is Google's Nano Banana Pro. Let's see this in action. First, we generate static image number one, a technical wireframe sketch of smart glasses. Next, we generate static image number two. The finished product shot of those same glasses, but now as a polished render with studio lighting. I then drop both images using the frames to video feature. Add a simple prompt like smooth transformation, static camera, and Google Flow fills in the motion. The wireframe materializes into the final product right before our eyes, which is the kind of thing that used to require expensive ass software. Now, you don't have to generate both images from scratch. In one of my previous videos, I literally took a stock photo of a chicken breast, used Nano Banana Pro to add a laser grid to it, giving me the before and after images, right? I then used Google Flow to animate the laser scan effect for me.

Now, some of you might be thinking, well, tools like Clling, Open Art, and Hicksfield do similar things. And you're right, but it's the same concern I raised with Whisper Flow. Third party tools can lose their edge overnight when big tech adds those features natively, right? For example, Google is already making Gmail AI native. So, I'm honestly not sure what that means for tools like Superhum. That said, Google kills products like it's a KPI. So, who knows? I could be completely wrong. So, let me know in the comments what you think. Check out part one of the series if you missed it. See you in the next video. In the meantime, have a great one.