Everyday AI Podcast – An AI and ChatGPT Podcast - Ep 716: ChatGPT’s new Deep Research Update: 5 Ways You Can Use it Today

Episode Date: February 18, 2026

Wish you had Jarvis-esque control over information? 🦸Now you do. While the AI drama and model releases were poppin, OpenAI snuck in a powerful update to its popular Deep Research mode. Missed it? You already KNOW we're giving it the 'AI at Work on Wednesdays' treatment to break down what's new and 5 ways you can use it today. ChatGPT’s new Deep Research Update: 5 Ways You Can Use it Today - An Everyday AI Chat with Jordan WilsonNewsletter: Sign up for our free daily newsletterMore on this Episode: Episode PageJoin the discussion on LinkedIn: Thoughts on this? Join the convo on LinkedIn and connect with other AI leaders.Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineupWebsite: YourEverydayAI.comEmail The Show: [e-mail] with Jordan on LinkedInTopics Covered in This Episode:OpenAI's Deep Research Platform Update OverviewNew Full-Screen Deep Research Report ViewerReal-Time Deep Research Steering and PausingUpgraded GPT-5.2 Model in Deep ResearchConnecting Apps and Targeting Company DataRestricting Deep Research to Selected WebsitesContext Stacking for Enhanced Research ResultsTop Five Deep Research Use Cases ExplainedTimestamps:00:00 "Revolutionizing AI-Driven Information Access"04:24 Enhanced Deep Research with GPT 5.207:41 "GPT-5 Pro's Game-Changing Features"13:14 "Start Here: AI Made Simple"16:03 "Context Stacking for Better Results"17:24 "Enhanced Site Research Tools"20:49 "Check and Reconnect App Connections"24:33 "Maximizing AI for Efficiency"27:19 "Optimized Internal Research Workflow"31:53 "Summarized Thoughts & Navigation"36:11 Personalized Competitor Analysis Tool38:45 "Organizing Gmail and Calendar Tasks"40:39 "Everyday AI: Thanks Y'all"Keywords: ChatGPT deep research update, OpenAI, deep research platform, ChatGPT 5.2, GPT-5.2, advanced AI research, latest AI model, research agent, app connectors, web sources, company data analysis, data-driven insights, business intelligence, citation checking, split screen report viewer, enhanced document viewer, live steering, model upgrade, personalized research, AI-powered productivity, automated research assistant, information retrieval, app integrations, Salesforce connector, HubSpot integration, ClickUp managementSend Everyday AI and Jordan a text message. (We can't reply back unless you leave contact info) Start Here ▶️Not sure where to start when it comes to AI? Start with our Start Here Series. You can listen to the first drop -- Episode 691 -- or get free access to our Inner Cricle community and all episodes: StartHereSeries.com Also, here's a link to the entire series on a Spotify playlist. 

This is the Everyday AI Show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life. Meet Firefly AI Assistant, now live in Adobe Firefly, the All In One Creative AI Studio. Just describe what you want to create and the assistant handles the rest, orchestrating multi-step workflows across Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome. The assistant accelerates execution. Imagine having complete control over your company's information in the web.

Let's be honest, whether you're an AI native worker or not when it comes to retrieving the correct information, whether it's from your company's files and folders or cloud or the internet, it takes a long time. But what if you had like a Jarvis-esque-enabled tool? that could find high-value insights from AI models without all of that nonsensical back and forth or those pesky hallucinations that come with using AI models. So while deep research tools have given us glimpses of this type of informational control that you really only see in superhero movies, I think even the most powerful options have left a bit to be desired.

Well, that may have changed this past week with Open AIs latest update to its deep research platform. And if I'm being honest, I think a lot of people missed it because of everything that was going on in the AI world. I mean, Microsoft was like, oh, we're not going to use open AI's models anymore. And Open AI and Anthropic are kind of fighting in this open cloth thing. It's been crazy. So I think the majority of people missed this huge announcement from Open AI because they just kind of rolled it out as a little tweet. But it is a new deep research update. And we're going to be going over it today and the five ways that you can use it starting now. So I hope you're excited.

I am too. This is Everyday AI, but this is our putting AI to work on Wednesday's series. So for the past year or so, we've been doing this every single week, a practical and actionable hands-on walkthrough for a new AI tool or model. So let's get into it. And if you are brand new here. Well, welcome to Everyday AI. My name's Jordan Wilson and we do this every day, not just Wednesdays, Monday through Friday with the unedited, unscripted daily live stream podcast and free daily newsletter, helping business leaders like you. Yeah, sift through all of that information better, save time to grow your companies and your careers. So if that's where you try to do, starts here. But make sure you take it to the next level by going to our website at your

everyday AI.com. Sign up for the free daily newsletter. We're going to be recapping the highlights from today's show as well as all of the important AI news today that is going to impact your company tomorrow and in the future. All right. Speaking of impacting you and your company now and in the future, if you haven't already, I cannot tell you enough times. Make sure you go listen to episode 712 and 713. That is our 2026 AI predictions and roadmap series. Trust me, It is literally thousands of hours of conversations over the past year, all into two episodes that you can't miss. All right, but let's talk about what we're here for. The new Chad GBT updated deep research.

And, you know, apologies to our live stream audience. If you're listening on the podcast, you didn't know this. Sometimes when I say unedited, unscripted, yeah, tried to do this earlier today. Audio went haywire. So sorry about that. This is take two. But regardless, let's get in and learn about the new deep. So what we're going to be going over out today show, the single upgrade.

One small little thing that I think turns now chat GPT's deep research from chatbot into a true research platform. Why your company data just became deep research's most powerful source. And last but not least, I'm going to be going over five practical use cases that can replace hours of manual research that you can do today. All right. So here's what's new. Open AI announced this last week with a tweet. Nothing else, no live stream, no big hoopla. And I think it's pretty big.

So it is available right now for Plus and Pro users. And I think it's going to be rolling out here soon to free users on chat. So here's the biggest updates. Visually, you'll notice it right away if you're using the new version of deep research. And I'll show you how to choose between the two. But reports now appear in a full screen viewer with a split. screen citation checking, as well as a nice little table of contents on the left side. So visually, it looks better.

It's easier to work with. And it's just a better experience reading these research reports. So there's also you can upload files both in the beginning and during the course of the deep research without interrupting it, which is huge. And then using that as your primary context for research. The other thing kind of related to that is live steering lets you pause or redirect the deep research agent logic in real time without having to start over. And then the one thing, this one little feature that I think is actually turning deep research

from a nice little tool to complete research platform is the ability to choose which websites it does go to, which is big. And then the most obvious update, but probably the biggest one there is the updated model. So now this is running GPT 5.2. So this is Open AIs, technically, their latest series of models available in chat GPT. There is a GPT53 codex, but that's only available in their codex platform, which, by the way, codex is absolutely insanely good, even for, you know, non-technical, non-coding work. Anyways, this new deep research is powered by their latest model available in chat GBT,

which is the GPT52 family of models. So unfortunately, we don't know exactly what flavor or very, that they're actually using for this. We just know it's GBT 52. I assume it's probably GPT52 thinking. I don't know if it's the GPT52 pro model. Hopefully, OpenAI will release a little bit more information, but for my use case, my testing,

obviously using chat GPT way too much every single day. My thought is it's an extended thinking version in not the pro version. Speaking of that, limits. Who gets it? How much? So if you are a pro user on that $200 a month plan like I am, you get a 125 full model and then 125 lightweight queries. So again, you know, like the last version of deep research was technically powered by a dual model approach.

O3 and O4. So those models aren't really used anymore. So I believe it was the O3 full and then the O4 mini. So presumably there's two different versions of 5-2. So once you hit your queries on the full version, then you get the lightweight. So yeah, hopefully opening eye releases a little bit more information on that because I think it's super important.

If you are on the normal $20 a month or a team's plan, you get 10 full model runs and then 15 lightweights for every 30 days. So, you know, essentially you get one every workday of the month between the heavy and the light. And then free users are expected to get limited queries in late February 2026. So I've checked my different free accounts. Don't have access to it yet. Obviously, I have access on my pro account, my Plus account, my team's accounts, all my other accounts.

All right. Here's why I think it's no longer just a mode. And now this is a fully deep research platform. Aside from the end goal is much better. The model is exponentially better. Right. And not just that, but the ability to pause, redirect and interrupt this model in the middle is such a huge game-changing option or feature.

So if you are a power chat GPT user, you'll know that actually opening I rolled out this feature, I think a couple of months ago, to pro users. So if you're using GPT-52 Pro, which is probably my favorite model, I do have to give Gemini 3 deep think a little more, a little more time to see if that can kind of take the crown in at least in my personal usage, right?

But with GPD 52 Pro, it can often take 20, 30 minutes, an hour longer, right? So deep research queries, if you haven't used it, the new model is actually a little slower, which I think is not a bad thing than the previous deep research. But deep research query might run anywhere from, you know, eight minutes to 45 minutes, right? It really just depends on, number one, what you're asking it. Number two, the data that you're giving it, the complexity of your query, you know, any steering that you do in the middle of the query. So it really depends.

And you know, I'll say this. If you're not running, whether it's on Chad GPT, Google Gemini, their new deep research power by Gemini 3 Pro is absolutely bonkers good. Right. Claude's research tool, not deep research, their research tool, perplexity, whatever. If you're not using a deep research tool, date, in connecting your data on the front end, I'm telling you, you are absolutely missing out. It is the best way to consume synthesized, well-sourced information at scale that is personalized

for you, your use case, your business's viewpoint, et cetera. It is literally, sometimes, right, I've worked with consultancies in the past, right? some deep research queries, if you give them enough information, enough context in that context window, I mean, it is like you hired a consulting company. If you do it right, right, it is an amazing output. So let's talk a little bit about why the 5-2 model matters. Well, it's enhanced reasoning capabilities for complex multi-source research tasks, just smarter research planning and improve synthesis across multiple sources. And you can still use the old model if you want to, but for the For the most part, there is actually one caveat that I'll share here in a little bit.

But for the most part, I do think that users are going to get a much better experience from the new 5-2 model. The document viewer is also great. So I am going to show that, right? It's our putting AI to work at Wednesday. So I will grab the screen here in a little bit and do some live walkthroughs. What could go wrong, right? Aside from my audio not working, like the first attempt today.

But research reports are now open in a dedicated full-screen interface. So it's just a nicer way to consume the information. It's less cluttered. The table of contents on the left is really cool. I like that. It's great to have. And then you also have the dedicated source panel on the right for easier and faster fact checking. And then the great thing is, is that real time control.

So you can obviously, like when you're using a thinking model, you can kind of monitor the chain of thought, but it's kind of a two-pronged approach. And I can show you. So it's going to both show you what sources it's looking at. It's going to, technically, there's three different things that you can see by kind of watching this in real time. There's sources it looks at, which is usually hundreds. There's sources that it will use, which is usually dozens. And then you can also see kind of the steps or the thinking that the model takes.

So it's great to be able to see that a little more granular control with this new version of deep research versus the last version. And then being able to adjust deep research in the middle with new sources or follow-up instructions midway like you could with GPT5-2 Pro. it's just such, it is huge, right? Even if you don't have the 30 minutes to sit around and watch the computer screen, if I'm being honest, schedule some time that you can with a meaningful deep research query, right? Especially if you're, you know, on the normal paid plan where it's somewhat limited. If you do that earlier on in the process, it's going to pay its dividends later.

I kid you not, one of the easiest things to do to get better results out of any large language model is to watch the chain of thought, you know, write down notes as it goes along, look at the output, compare your input, your notes as you go along in the output, and then run it again, right? Such an easy shortcut to get much better. And I think with a new version of deep research, aside from the fact that you can interrupt it midway, always running a second or third time is going to give you better results. All right.

So now let's do this live because I do want to show everyone here a little bit on kind of the new setup, the new layout. So I'm going to share my screen here, live stream audience. Thank you for letting me know earlier that my audio wasn't working. So let's try this again. What could go wrong, right? All right. So let's bring up my chat jvety window. Adobe just introduced an entirely new way to create, bringing the power and precision of its creative suite into one conversational experience. Meet Firefly AI assistant. Now live in the Adobe Firefly app, the all-on-end in one creative AI studio. Powered by Adobe's creative agent, Firefly AI assistant lets you start with your vision,

just describe what you want, and shape the outcome as it takes form with the assistant. The assistant orchestrates multi-step workflows, drawing on 60 plus pro-grade tools across Adobe Creative Cloud apps, including Photoshop, Illustrator, Premiere, Lightroom Express, and more to help bring your ideas to life. You can also get started with creative skills, a growing library of pre-built workflows for common creative tasks, like batch editing photos, creating mood boards, portrait retouching, and creating social variations. Every step the assistant takes is visible so you can refine, redirect, or take over at any

time. You stay in the driver's seat as the creative director. Adobe Firefly AI assistant now in public beta. See it today at firefly.adopi.com. All right, there we go. And, you know, podcast audience, I'm going to do my best to describe this, but if you want see the video version. You can always do that on our website at Your EverydayAI.com. You can always listen to the podcast there as well, each episode page, listen to the video version, et cetera.

Okay. So let's go ahead. I'm going to clear my little computer interface out a little bit here. Okay. So I just started a deep research query, but I'm going to walk you through how this new version works. So right now I am on my, looks like I'm on my team's plan because I have this company knowledge button, which is a little different than if you're on a normal plan. It's not super important. But so now to start a deep research query, you'll see kind of what I'm doing here. You're going to look in the prompt bar, click the plus button. Okay.

You're going to see the deep research option in the menu that pops up. So here's the thing. A lot of people are overlooking this. Now there's a drop down menu after you select deep research. And from there, you can select the version, the new, updated version is just called deep research. The older version is called legacy. So why would you want to use that legacy version, right?

It's an old model. It's 03 and 04 mini, right? You want to use 5-2 for the most part. There's all these new features that you just told me about Jordan. Why would you ever use legacy? I'm being honest, there is one little thing I actually like better. So in the old model, so in the legacy model, right, before it got started, before the deep research got started, it would usually,

ask you three to five clarifying questions, okay, which is always nice because what if you, you know, just type something that's nonsensical, if you make a mistake, right? You don't want to have to wait in the old version a long time for it to be done. So I like that it would ask follow-up questions. So in the new version, there is a similar feature. Essentially, it puts the plan together and you just have to approve it, right? Technically, I like the old version better because those questions that it would ask you, really, I think, lead to a better first version. So what I would recommend, and if you've taken our free Prime Prompt Polish course, you know, which is just updated, it's free inside of our inner circle community, you know this

concept of context stacking. I would context stack first before starting a deep research query, FYI, two to three times better results easily. All right. So now let's go ahead and jump in So I have a prompt up on my screen, a deep research query, and it's already working. So I'm going to show you what's happening. So I essentially said I use Canva every single day for building slides for the everyday AI show. Please do not look at these titles, as sometimes I do not always update the titles. So essentially in my deep research query, you can choose the different apps that it has access to. Right.

Last time I checked, there's 60 some different apps that you can connect your data. to. So I use Canva every single day for my, you know, ugly slides that I put up on the screen here on the live stream. So I have 720 some Canva decks that have a wealth of information, right? And then I'm also giving deeper research access to my website. So this new feature, you can click on manage sites. All right. And you can choose a specific site, which is great. And then there's also a new toggle option. So I have it on my screen here, but it's very easy to see this. And then it says prioritize these sites, but allow full web search.

So essentially, you can require deep research to go to your site first or, you know, a series of sites that you trust, your competitor sites, etc. But you don't have to limit it, but you can if you want. All right. So in this use case, I'm actually going to do that. All right. And then it gave me a kind of a research plan here. So it says, I'm going to extract live stream decks from the user's Canva account using the Canva connector.

It's funny that OpenAI is still calling it a connector, even though it's not what it's called anymore. It's called the app. Anyways, let me get back to my original prompt. Sorry, got sidetracked here. So I said, please look through the last six months of Canva documents that appear to be live stream presentations for the everyday AI show. This is important, all right, because I obviously use Canva for dozens of other projects. I'm saying ignore those, right? The names are all over the place.

So go in, use your best judgment, smart model with computer vision, and find those that look like that they're live stream documents. And then I'm saying cross-reference those Canba Decks with the web pages on my website, Your EverydayAI.com. Then create an easy to digest report that goes over the 50 most popular trends, categories, stories, news, happenings, events, LLM updates, new AI models, etc. This should be angled as a starting guide for someone who is newer to AI, but who wants to double down on their knowledge. So even the last 50, right, I said six months, it's like 150 episodes. That's a ton. Even myself, I've probably forgot 80% of what was covered here. So this is hopefully going to be a good guide.

But now you see deep research kind of creates this plan. So it's a couple steps. It gives me the different steps that it's going to do. It's going to extract information, then it's going to cross-reference. it with the website. It's going to survey additional high quality web sources. It's going to identify and rank the top 50 trends. And then it's going to draft an easy to digest starter guide. So if I want to update that at any time, well, there is an update button. Then essentially, what's going to happen here? I can click to add files or I can just essentially send a follow-up prompt. All right. So I'm not

going to do that because we're going to give it some time to cook. But now you see how this works in real time. You see some of the new features already. And we're going to check in this at the end and kind of see some of the other features that are on the back end once a report has been produced. But you can see it's going. I can also click. I wish OpenAI made this a little more prominent. People don't know. But there's essentially this small little gray text at the bottom.

So if you click on that, that's how you watch. that's how you watch it work live right so now i can see its research activity i can see the steps that it's going through this is important because sometimes maybe one of your apps that you have connected maybe the connection is stale and you need to go reconnect it and you think it's connected but it's not right so essentially especially if you're using a lot of apps you always want to keep an eye early on right because you don't want to come back 45 minutes later and like how Frick, right? My, I changed my password two weeks ago and forgot to update the app. And, you know, a lot of times deep research and AI models, when that does happen, they'll try, you know,

crazy things to try to make up for not having access to certain information that you told it it had access to. And it'll try for sometimes way too long. All right. We'll check back in on that later. So let's go ahead and get back to learning. All right. So like I said, you can connect the apps and target specific websites. And this is huge. All right. So I'm going to give you my use cases here in a minute. And I've already started to.

But this is where I need you to think. What is it you do? I think so many knowledge workers out there, that's probably most of people listening to this podcast. If you're sitting in front of a computer all day, you know, you're probably using different apps, different pieces of software, right? As an example, maybe you're using Salesforce as your CRN. Maybe you're using HubSpot for email marketing.

Maybe you're using, I don't know, click up for project management, right? All of those that I just mentioned, they all have connectors, right? Maybe you use, you know, SharePoint in OneDrive or I don't know, Gmail, Google Calendar. All of those things have connectors. So what do we as knowledge workers do? Well, we go visit Salesforce. We go look at Slack. we go look inside HubSpot at our last campaigns, right?

We go check the project and click up. We go, you know, log into our, you know, our OneDrive, our SharePoint to check these files and folders. All of those things that we do. Large language models, especially ones that are extremely powerful, like this new deep research from Open AI, they do it better, they do it faster, they do it at scale. I don't care. Better than me, better than anyone. This is what, right?

This is what I've been doing for 20 plus years of my professional career. Use different software. Go find the information, right? Essentially, you are synthesizing, personalizing, and carrying context over from app to app. Maybe you're taking notes. Maybe you're working on a document as you go along. But that's what we do, right?

That's all we do. But now this is what deep research does, right? And it's not just Open AI's version, right? Anthropics version, very similar. Google's version, very similar. Perplexity's version, very similar. They all have different apps or connectors. But one thing to keep in mind, which is very different than an agent, right?

Because technically, at least when Open AI announced this, they said it's a deep research agent. Deep research, this only has read access. So it's not going to perform actions for you. Obviously, with their agent mode, you can do that if you want to, and still using a lot those apps, it's much slower. So deep research is not going to, you know, delete those files, right, in your CRM, or it's not going to change the status of an important project in your project management tool. It is read only never write actions. That's important. All right. So let's go over now, now that you know how it works and we'll check in on our little project to see if it worked.

Now let's go over the use cases. So I have five that I think are great. And I already gave you kind of my, you know, an example of my use case. But as I go over these five use cases, I want you to think about your work. Where do you spend your time? Even as you are using AI tools, right? Where are those inefficiency still? I think with deep research, a lot of those are going to go away.

I think it's an underused, aside from canvas modes, right? In open AIs, Chad GBT, GBT, Canvas mode in Google Gemma, or the artifacts mode in Claude, right? I think Canvas modes or artifacts is underuse and deep research is underused. And the biggest thing, right, is when you select those apps that you want to use or the websites, you can select multiple. You can select 10. These are the 10 apps that I use every day.

Again, assuming you have permission to connect all these apps to your Chad GPT account, right? Always do that first. That's what we do all day. We carry information from app to app website to website. And well, we create something in the end. So this is where it's huge. So use case number one is a memory powered planning for your next steps.

Sometimes I like being very open-ended with chat GPT in terms of what you're working on. And this becomes especially powerful as you give access to chat GPT to more information about you, about your goals, about your team, what you're working on, your company, etc. So this is if you have the personalization term. on, the memory turned on, right? This is great. And you'd be surprised. So my example, and I invite you to try something like this, just say based on everything you know about me, including memory, chat history, et cetera, please plan out my next six months of what I should be focusing on, in my case, to grow everyday AI, right? Start open-ended. Don't give it access to everything else,

right? Start open-ended, and then you can do a follow-up prompt if you want to based on what it suggests, and then give it access to certain information. I think, One of the biggest mistakes people make when working with extremely powerful large language models is we think that we know the right answer. I always say, say, start, wide, work your way to narrow. Okay. It's, you're going to find out some great insights because one thing that I always say, especially if you're a power user, this is why I think advertising on the chat chb t

platform is going to be bananas good. They find gaps that you don't even know about, right? If you're asking about A, B, C, D, E, F, right, a large language model is going to be able to connect patterns across things that you may not even know about yourself personally, professionally, career-wise, your team, et cetera, right? Because it understands the intent of what you're asking over and over. It's going to be able to spot patterns that you used to ask about, but no longer do. You know, it's going to connect these dots that you may not necessarily even know are there to connect. So start lied. Use case to essentially rag company search, right? This is not, again,

This is not as good as a complete fully fledged vector database, but this is huge. And I think that, you know, using the deep research mode is going to give you a much more accurate and better cited report than using a normal thinking model and then using apps that way. So in this case, I say restrict deep research to search only your whatever it is, you know, your Google Drive and your company website as an example. And then the connected document stores just become. these searchable sources for focused internal research. And then you can get a synthesized report built entirely from your

organization's own files and your files only. So my example of this, well, that's what I just showed you, right? I can't tell you, if I'm being honest, how valuable I think my Canva account is and our website, right? There's a lot of inaccurate information out there when it comes to covering the AI. And that's why I love this new sites feature as well. And you can include sites because I know. as an example, you know, there's a lot of sites out there, web publications that you think,

oh, by looking at their name, but then there's different versions of those, right, different countries, different publications under that umbrella. I know at least a dozen that you would think would be very reputable. I know that they just turn out AI slop. It's full of hallucinations. Anytime I'm doing a normal search, I say, oh, can't include that, right? I do hope that Open AI allows you to insist. of include websites one by one.

I hope it allows you to exclude or blacklist websites one by one. That would make it a lot better. I did suggest that, you know, on Twitter and they actually liked the reply, which they normally, you know, don't go through and like replies. So maybe that means it's coming. Maybe it means nothing. But that's a great example is, you know, give it access to all of your company's data, only your company's website.

And that's it. Go to town. Right? This is a very quick version, not as good. Again, is a full retrieval. I'm going to generation set up, right? But it's 80% of the way there and one percent of the time. The other thing that's great or something that's important to understand and know,

where does data come from? There's three sources, training data, right? And we don't necessarily have control over that. We can kind of prompt our way around it. So large language models have training data. then there's number two, the data that we connect, whether that's through, you know, apps formerly known as connectors. It's something we upload in a chat window, a project file, et cetera.

Then there's three websites. So training data, data we connect, and websites. So this is a great way to control the latter two by just restricting them. And then you have a version of RAG company search. All right. So let's go ahead. Let's see how ours did. So yeah, unfortunately.

it's still it's still working all right i wanted to be able to see if we could see some results here luckily like any good um you know person trying to cook in the kitchen with some ai stuff i do have a version of this done so yeah uh the example i gave that was kind of the you know everyday AI retrieval i've met in generation right just hey only look at our wealth of canva decks and our website it's still going uh right still going it's been going for uh quite a bit here and I can kind of check on it, still cooking. But let's go ahead. Let's look at a finish version here because we do have a finish version, right?

I put one cake in before we start it. So this did take 35 minutes, which is probably one reason why we couldn't get it done in the live stream. Sometimes I've had ones that take, you know, 40 minutes. And then the second time, it takes 20 minutes. So I was giving it a try. But anyways, let's look now live at some of the new features and options. So there is this new kind of full screen view here. And then on the upper right hand side of the screen, right?

So hopefully podcast audience, simple enough to follow along. So when the deep research is done, you're going to click on it. It's going to put it into live, sorry, full screen mode. In the upper right hand corner, you can download it. You can copy the contents. There's this little squiggly line. If you click that, that's how you get your sources. So these are the sources used.

Okay. And then you can click the activity, which is it's kind of summarized chain of thought. This is how it thought about things, went through the sources. And this to me is fascinating. I spend way too much time reading summarized chain of thought more than probably most humans. And then on the left hand side, this is the new table of contents, which is really cool. So especially if you have longer reports, you can just hover over these little toggles here on the screen. And then you can scroll down.

And then if you want to click something, it's like a jump link. So I can click that and it goes straight there. So actually, let me see. Let's see how this turned out. Okay. So we have an executive. summary. Let's see if it did everything we asked. Okay. So this is the great thing. It's giving sources. So this is meta. All right. There's something about deep research that just came up from my website

in the Deep Research report. So I can click it. And then on the right hand side, it has all the sources that are used. So if there's ever anything that you want to verify, you can always click that, look at the sources and verify it. All right. This is not your initial chat GPT hallucination, right? 2020. This is not it. Always cited and sourced very well. All right.

So let's go down. Let's see how it did according to the directions. I gave it. All right. So scope, sources, and cross-referencing. So it's telling me what it did, how it worked. All right. It kind of mapped out all the different cambod decks.

That's really cool too. So I can go check those out if I ever have any questions. Okay. This is nice. It gave me a visual synthesis of the six-month landscape. I didn't even ask about this. This is pretty cool. It made me kind of a visual of how things have changed over the last six months

based on all of the data, which is very helpful. All right. It gave me a kind of what themes were talked about the most. Unfortunately, the X access is a little messed up and it's overlapping. But if I wanted to, I could look at the code and kind of figure out what it's doing. So, you know, I didn't even ask it to create these. visuals and it actually did a really good job of making this document more interactive, right? It made some charts.

Here's the ranked top 50 items with side ready briefs. Sorry, slide ready briefs. Okay. This is, this is very cool. So we found the 50 trends that have, you know, the 50 biggest trends over the last six months, at least according to things that we've been talking about here on the website. This to me is such a valuable resource, right? Imagine doing this for your company's website, for a competitor's website, for 10 industry websites that you follow all the time, right?

Maybe you've been out of the loop, been busy on a couple projects, been on vacation, right? This is such a good way to do some simple sentiment analysis and catch up quickly. So it gave me kind of the recurrence signal strength, you know, things that were very high. So AI as an operating system, that's a trend, the number one trend. AI agents becoming the default workflow units, very high trend. So this is really cool. So this is really good. Okay.

And then we have slide ready briefs for ranks one through 50. So it gave me a nice write-up on all 50 of those. Sheesh. This is good. This is nice. My gosh. Okay. I'm going to read this.

Right. As crazy as it sounds, like I told you, I don't remember everything I've talked about over the past six months. You know, sometimes, I don't know, I feel humans. maybe it's because of AI or just attention spans in the internet. But sometimes I feel like, you know, we have like a memory of a goldfish. This report to me is pretty freaking amazing. So yeah, I don't know.

If you want access to the report, just go ahead, repost today's show on LinkedIn. And I'll send you this because as I'm scrolling through, this is actually very impressive and a very good resource to go through and read. So yeah, if you're listening on the podcast, we always have. the linked in link that you can just go and then repost this and I'll set it to you. Really cool. All right. Let's wrap up. Let's go over our next use cases.

So use case three, a competitor deep dive using your company context. So similarly, how I started the first use case by saying, hey, use everything you know about me. Same thing. Use everything you know about my company. But then also combine your personal context, what your role is. And then your synced company data for competitor analysis. Then you can add your competitors, websites, industry, web, websites, et cetera. And then you can get a report that's mapping competitors against what matters to you

and your specific company. Again, this is like as if you were hiring a consultant, but for you, right? Not for your company or for your department. And if you do take the time and even if you go through maybe two or three iterations of providing feedback or steering it midway through, I think you will literally be shocked at the amount of information that you get out. Have you guys never done this? Sometimes I feel like I'm a crazy man. Because I'm like, why is no one talking about this and doing it nonstop. It is so good, right? I can't lie.

I think a good chunk of whatever success we've been able to have as a, you know, as a top AI podcast. And I know that's kind of cringe to say. I think a lot of it is from doing things like this, things that I'm telling you to, right? When I talk about these use cases, these aren't just like random things I read about on the internet. I'm like, oh, this is cool. Go try it.

These are things that I do all the time. And then like when I do them, I'm like, holy, Frick, that's amazing. I need to tell people, right? So you should be doing this. Use case four. Industry SWAT built from your data and news. So kind of similar to three, but more on the industry versus just competitors and just a SWAT, right?

Strength, weakness, opportunity threat report. So this can use your chat history and company data to identify sector trends that are impacting you. Deep research can scour trending industry news, filtered through your business context, and then produce a SWAT analysis grounded in both your data and current market signals. The cool thing, and I do have to do a little bit more research on this, but I'm pretty sure deep research does a great job with Boolean URLs, which is amazing because without being too dorky, what you can accomplish with some simple URL hacking, right? You can essentially have a dedicated, up-to-date research assistant that you don't have to keep feeding, you know, different websites, right? If you know what you're doing around some simple Google search operators, yeah, it's extremely powerful.

Last but not least, or at least I do that all the time with GBT 52 Pro. And I have tested it a little bit in the new deep research, but I got to do a little more testing to see if it is consistently handling that. And then last but not least, use case five, the follow-up system that scours your inbox and calendar. This is huge. I miss so many opportunities, so many emails. I stink at it, mainly because I get spam.

So my example here, I said, please carefully comb through the last six months of the connected Gmail inbox in Google Calendar. For Gmail, pay specific attention to my Outbox as my inbox gets spammed a lot and the good majority of what lands in my inbox is not important. However, if I have replied to something via my Outbox, that means it is generally important. For my Google Calendar, please look to see which meetings I've had with other people in the last six months and cross-reference that with correspondence in my Gmail inbox. The goal is to both follow up on opportunities for everyday AI, where I may have dropped the ball or forgot to respond, as well as to reengage older conversations that may have already closed in theory, but may be worth revisiting. Please keep in mind all the context that you know about me, as well as looking at the two attached informational sheets. And what I've attached here is I essentially have these living, breathing, markdown files that I always update anytime I'm working in really any large language model, both about everyday AI and about my role, kind of, you know, my day-to-day. what it looks like.

So I have two different markdown files about everyday AI and then about myself. So I mean, y'all, I should actually spend way more time on this because there's no reason for me to suck at email. It's just more or less overwhelming. Right. When I run these, it's like here's, you know, 85 extremely important emails that you haven't got to. Right.

Unfortunately, they can't send or draft replies yet. But hey, maybe one day. All right. So that is a wrap. So now you know what is new in. Open AI's new updated deep research and five ways that you can use it today. So yeah, if you want to go check out that trend report, make sure to go share this and repost this on LinkedIn.

And then 712, 713. Don't forget those numbers. That is the 2026 AI predictions and roadmap series. If you haven't listened to those, you have to. And then please go to your everyday AI.com. Sign up for the free daily newsletter. Thanks for tuning in for putting AI to work at Wednesday. Hope to see you back tomorrow and every day for more everyday AI.

Thanks, y'all. Meet Firefly AI Assistant. Now live in Adobe Firefly, the Allman One Creative AI Studio. Just describe what you want to create in your own words and the assistant handles the rest, orchestrating multi-step workflows across Adobe Creative Cloud apps, including Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome while the assistant accelerates execution. Stand control with the ability to step in and refine at any time.

See it today at firefly.adobie.com. And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going. For a little more AI magic, visit Your EverydayAI.com and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.