Everyday AI Podcast – An AI and ChatGPT Podcast - EP 612: ChatGPT Connectors: What they are and why you NEED to rely on them daily

Episode Date: September 17, 2025

Remember RAG? 🤔When companies would spend multiple six or seven figures (and sometimes a year or more) trying to connect their enterprise data to LLMs?Now there's an easier way. ChatGPT connectors. In a few seconds, you can get like 80% of RAG's power for like free 99. Yet... so few people actually use them. Let's change that and put AI to Work this Wednesday. ChatGPT Connectors: What they are and why you NEED to rely on them daily -- 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 today's LinkedIn stream and connect with other AI leadersUpcoming Episodes: Check out the upcoming Everyday AI Livestream lineupWebsite: YourEverydayAI.comEmail The Show: [e-mail] with Jordan on LinkedInTopics Covered in This Episode:ChatGPT Connectors Overview & BasicsConnectors vs. Traditional RAG ComparisonSetting Up and Using ChatGPT ConnectorsSupported ChatGPT Connector Integrations ListChatGPT Connectors Security and PermissionsModes: Chat, Deep Research, Agent, SyncedReal-Time Data Access with ConnectorsMulti-Connector Workflow and Productivity TipsBusiness Use Cases for ConnectorsConnectors vs. RAG: Cost and DeploymentModel Context Protocol (MCP) & Custom ConnectorsTime-Saving Connector Examples and DemonstrationsConnector Best Practices for EnterprisesTimestamps:00:00 "AI Weekly Workflow Overview"04:31 "ChatGPT Data Connectors Explained"09:05 Efficient App Connectors Overview10:12 "Enhancing Outputs with Language Models"13:14 "Gmail and Google Calendar Sync"18:36 "Setting Up AI Connectors Protocol"21:57 "GPT-5: Versatile Data Integration"24:05 "Optimize Connector Usage"27:37 Organizing and Formatting Information32:03 "Agentic Model Streamlines Document Tasks"34:05 "Optimizing HubSpot with GPT Connector"39:08 AI-Powered Task Automation Benefits41:16 "Maximize AI with ChatGPT Connectors"43:04 "Maximize AI with Data Connectors"Keywords:ChatGPT connectors, ChatGPT connector, connectors for ChatGPT, AI connectors, custom connectors, model context protocol, MCP, secSend 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. Remember when RAG was all the rage in AI? I'm talking about retrieval augmented generation,

essentially when many companies would spend multiple six or seven figures and many times half a year, a year or more of development time, just to connect their company's proprietary data to large language models. Yeah, it's a little different than what a chat, GPT connector is, but it essentially does that for as little as $20 a month in about 20 seconds, you can get a good portion of what many enterprise companies spent, like I said, countless dollars and countless hours on. Yet, I'm surprised.

I do a lot of consulting for a lot of enterprise companies, you know, those that are investing heavily in chat GPT enterprise licenses and they need to train thousands of employees. and yet so many of them still aren't even using these chat GPD connectors. So that's why I think it's time to go over what the heck they are, how they work, and well, why you need to actually be relying on them daily, not just using them. All right, I'm excited for today's show. I hope you are too. What's going on, y'all?

My name is Jordan Wilson and welcome to Everyday AI. This is your daily live stream podcast and free daily newsletter helping everyday business leaders like you and me, not just keep up with all these AI developments, but how we can make sense of them and put them to work for us to grow our companies and our careers. If that's you, like, hey, that's what I'm trying to do, Jordan. Welcome. This is your new home. So it starts here on the live stream podcast, unedited, unscripted.

But if you want to take it to the next level, that happens on our website, your everyday AI.com. Go there, sign up for the free daily newsletter. We're going to be recapping today's episode as well as all the AI news. You need to stay up to date. Also, you can go watch literally 600 plus. of our backlog of episodes, listen to them, read them, all on our website. It's a free generative AI university, so make sure you go check it out.

Enough to chat, let's get straight into chat GPT connectors, what they are, and why you literally need to be relying on them every time you open chat chvety. So in today's show, we're going to detail the basics and also advanced functionality of chat chvety connectors. Talk about the similarities and differences between connectors and traditional rag or retrieval augmented generation. I'm going to show you live. Yeah, we're going to begin to work live here on how they work and how we use them,

how I personally use them as well. And I'm going to give you three or maybe more real use cases. All right. So this is part of our new AI at work on Wednesday's segments. So essentially here's what we do, Monday through Friday. Monday, we bring you the AI news that matters. Tuesday, it's a hot take Tuesday. My take on something happening in the world of AI on Wednesdays,

we put AI to work, show you a. new tool, technique, mode that was released, usually from the big companies and how we're doing something. And then usually Thursday and Fridays, we do different interviews or I might just do another show on something timely and relevant. So let's put it to work. Well, actually, if you really want to put it to work, make sure you go repost this episode. Y'all, I've been having so much fun, because yes, I know sometimes these episodes go a little long. I'm trying to keep them to 30 minutes, but they always creep up to like 35 or 40 minutes. But I have so much extra information in great material that never ends up making the cut for what I'm going over in the show.

So I've been putting together a lot of these kind of companion guides. And this one is amazing. There's so much additional information. I just didn't have time to cover today. So make sure you go repost this show on LinkedIn. And I will send you the chat GPT connectors cheat sheet. It is live. It is ready to go.

So if you're listening on the podcast, I always put the link to the LinkedIn post. this actual live stream, go repost this and I will share that with you. Also, some other recent chat GPUT tutorials and guides that are going to help you. Go listen to episode 599, the five new overlooked chat GPD features you should be using but aren't. And then episode 588, chat GPD's updated canvas mode in GPT5. What's new and how to make it work for you. All right.

So let's talk about chat GPD connectors. So essentially think of these as secure bridges and they link chat GBT to, well, your data. specifically within Google Drive, your Outlook, SharePoint, HubSpot, and more, we're going to be going over all the different connectors as well as how custom connectors in the model context protocol work. But essentially, you might be figuring like why. Couldn't I just, you know, while I'm using chat, GPT, couldn't I just go fetch information from the internet or individually upload files?

Well, yeah, you could. But this is a way to essentially bring your data into chat. Chepti. This is a version. I like to say this is kind of like mini rag. This is a way that you can right. So instead of chat chitpt going and searching for answers aimlessly inside of its training data, which is just the entire internet gobbled up, chewed up and spit out right back at you, sometimes good answers, sometimes not. Yeah, you can manually go search for that file or you can have chat chitp t go to a certain website to hopefully give you better responses and a higher quality output or you can use connectors and connect your data. So I do have to always put this

thing out here. Big asterisk, right? You know, make sure that you have permission within your company to update all of this. It's very similar to, you know, if you're using, you know, Google Docs and there's a third party extension that you need to use and it's going to be able to view all of your information inside of your account. It's the same way. Right. So we're going to be going over a little bit more on the security. That's essentially how it works. And what this gives you now is real time answers out of chat jvt grounded in your business data not just general internet knowledge or just random training data which may or may not be helpful and this i think is one step to transform chat gvt into a true business assistant with citations you can trust so here's why not

using it can be a failure and why when i say you should be relying on this not just using it you to be relying on this because generic AI, right, can access your files, your information, your CRN, your calendar. So what this usually means, and I think in the 2023 and 2024 phase of chat GPT and other large language models, it was a lot of copy and pacing, right? When we talk about context engineering, right? Another hot and trendy term, just like RAG was in 2023. This is essentially a shortcut to context.

engineering because it's going to bring in automatically all of your business context once you select these connectors. And right now, business leaders are wasting countless times, right? Especially power users. I'm shocked, right? When companies hire us to help them with Chad GPT training, you know, front end AI strategy, I'm always shocked how few people, how few companies have their connectors properly set up or even using them. All right? Because what this means is they're just having to go search for files manually or go and create files and bring them in to give chat GPT better context before you start.

Because you know the more conversation, the more context that you share with chat GPT before you're going after a desired output, obviously the higher quality, hopefully more accurate and relevant that that output is going to be, the more that you work with it on the front end. So anyone that out of the, I don't know, 13 or 15,000 of you that took our prime prompt polish course, yeah, it's going to be coming back soon, I swear. Right? You know this, right? That's what we talk about in our refine Q method.

You have to make the model smaller, smarter, and more specific for your business needs. And that's exactly what connectors do. But without connectors, you're really just at risk, at higher risk for hallucinations or just giving generic, outdated, or incomplete answers. And the thing you have to think of as well, chat, GPT, just like any generative AI tool, any large language models, it's generative. So you may run the same prompt 10 times, get 10 very different answers, right? Sometimes it might automatically go on the internet.

Sometimes it might pull old data from 2022, the exact same prompt, right? That's why it's so important to be using these connectors and relying on them, especially if you're using chat GPT for business. So here's a little bit how they work. So you log in once using just, standard authentication like you would authenticate, you know, any SaaS product. And then chat GPT gains secure query access. So right now, there's not read, write permission within connectors.

That's something to keep in mind. But you can do that via MCP or custom connectors. And then connectors inherit your existing app permissions. So there's no unauthorized data exposure. And then we're going to go over all the different supported connectors here in a second. But there's different ways that they work, which is. important because previously a lot of these connectors only worked with deep research, which when they first came out, I'm like, okay, well, this is great, right?

But you might not want to wait 10, 12, 15 minutes to get something back. You might just want, you know, chat, Chb-T to go through your Google Drive account, you know, maybe take two or three minutes and fetch different files, different information from different files. And that's it. So when they first came out, I don't think they were super useful. Because at that point, right, if you had to wait 10 to 12 minutes, it's like, okay, what's the difference then of just, you know, not doing a deep research and then manually doing that context engineering or manually uploading a couple of files because then you can use the normal quote unquote chat mode where it's much more instantaneous, right?

Because in the end, we care about productivity. We care about getting higher quality outputs than the time that we're spending putting in, then we wouldn't or then we would be getting if we weren't using large language models. So, you know, a lot of times, I think early on, The ROI wasn't always there early on with connectors, but now that you can use them in normal chat mode, they are. So the three different ways you can use them is there's chat search, which is an instant file or email lookup with clickable sources for verification. You have deep research, right? So if you really want to get in depth, that can synthesize across multiple systems for complex multi-source analysis and then synced connectors as well.

So this essentially pre-indexed drives and repositories for lightning fast responses. So in some instances, it's having to go out and kind of quote unquote, crawl your information statically. In other instances with synced connectors, it's going to index all of that. And that's essentially where you get a version of rag, right? It's kind of like vectorized embeddings and, you know, your information's there. And it's going to kind of go against that embedding before it goes to the large language model. So like I said, three different modes of operas. So if you need quick live lookups, in a lot of the connectors, you can use them in different modes.

And I'm going to go over that here in a second. So if you just need something quick, just do the normal chat. If you need something more complex analysis across multiple files, across multiple connectors, you might want to activate deep research first and then choose your connectors. And then if you need synced files, you know, make sure that you use those options. All right. Here we go. We're going to quickly go over the different connectors as well as the modes that they work in.

Because actually didn't mention, you can also use them in agent mode as well, which is really cool. And I think a lot of people don't know about that and don't even use that. All right. And if you caught our agent show and if you shared that show, I gave you the secret on how to schedule agents as well, which a lot of people don't know about. That's why you got to go share these episodes, y'all repost them on LinkedIn. I give you all the secrets that people don't know.

All right. So we're going to go over these quick, rapid fire. I'm going to tell you the connector, what it can do and kind of what mode it can work in. Here we go. So, Box. You can search and reference files, and this works in chat,

deep research and agent mode. Canva. This can find and fetch your Canva designs. It's not going to, you know, again, it's not going to write to them or design anything, but this is available in chat. deep research and agent mode. GitHub.

So GitHub is one of those that is synced. So it will sync your repose. So this lets you access repositories, issues, and pull requests. And it is required for some features such as codex. If you're using chat, youpcate's codex. And that works in chat deep research and agent mode. Again, that is synced. Another synced option is Gmail.

This one, banger. I'm so glad that this is no longer in deep research. I was, you know, getting impatient. So with Gmail, you can find in reference emails from your inbox. It is live automatically. You don't even have to enable it once you, you know, you don't have to check it once you enable it, which is great. And that works in chat, deep research and agent mode. Similarly, Google Calendar. Once you connect it, it is used automatically. And it works in chat, deep research and agent mode. You have Google Drive. So Google Drive, you can search and reference files from your

drive that works in chat deep research and agent mode dropbox similarly find and access your stored files chat deep research and agent mode uh google contacts uh so this is something that it will recommend you can let chat gpt recommend connecting to google context when responding when appropriate this lets you reference saved contact details right if you're like hey what's bill from it's phone number and email right you can do that this is right now only available in chat uh not in any other modes. HubSpot. Here's low key. This one is good. I use HubSpot a ton for some other work, not necessarily for everyday AI, but this lets you reference contacts, deals, and CRM data. And this works in chat, deep research and agent mode. Linear. So this lets you find in reference

issues and projects. If your company uses, you know, especially if you're a dev team, you probably use linear. That works in deep research and agent mode. So not chat mode, just deep research and agent mode. And then notion. Notion is a newer one that was added a little more recently. This helps you search and reference your notion pages, works in chat, deep research and agent mode. Then we have the Microsoft lineup here. We have Outlook Calendar. This just lets you look of offense and availability. Unlike Google Calendar, this is not kind of enabled automatically, but this does work in deep research and agent mode. So again, no quick ones, no quick chat queries from Outlook.

Same thing with Outlook email. So Outlook email and Outlook Calendar. You don't get the quick responses. So Outlook email, same thing. Search and reference. Your Outlook email only works in Deep Research and Agent mode. But SharePoint, you get that instant response. So SharePoint does work with chat, deep research, and agent.

And that allows you to search and pull from shared sites in OneDrive as well. So OneDrive is not his own connector. It is actually working under SharePoint. And then last but not least, I think last, yep, Teams. So Teams lets you look up chats and messages. And luckily, this is available in all three modes, chat, deep research, and agent mode. I got to take a sip after all that. Live stream audience.

Let me know. Number one, have you used these connectors so far? And if so, or maybe if not, after I just read them all, What one are you looking forward to the most? And if you have any questions, you know, go ahead and get them in. If I don't get to them here in the live stream, I will make sure to get to them afterwards in the comments. All right. 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 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. Let's look live. What could go wrong?

Nothing. All right. Hey, live stream audience, if you could. me know if you have my screen. So what I'm going to do here, I'm actually going to get a, a prompt going right away, because I know this is going to take a little while. All right. And then I'm going to open up a new tab here.

And I'm going to show you around, I'm going to show you around connectors. Okay. So if you are brand new to chat chippy T, it's, it's going to look a little different depending on what plan that you're on. interfaces change a little bit. I am on a pro account. I'm going to resize my window here for our live stream audience. So hopefully you can see.

So I'm using Chad Chd-GPT thinking mode. So you also technically have a little bit of fine-tuned control, even when you're using just the chat mode, because you can use just the normal GBT-5 or you can use GP-T-5 thinking. Connectors do not work with chat-G-GP-T-5 pro. And also, I don't know why it shows that they work with thinking mini, but I've never gotten them to work. So for the most part, you can use chat GBT5 normal or you can use chat GPD5 thinking when you are working with connectors. Okay.

So what you're going to do, and again, your interface might be a little different, but you're going to click on the plus button in where you would normally set a props. Or you can click the backspace, or not the backspace, the, a backslash button and then it's going to bring up different options for different modes or tools that you can use. From there, you're going to go to use connectors. Okay. And then by default, I already have some connectors connected, right? And then you are going to choose your sources. But if you don't have any connected, all right, you're going to click that sources. And then you're going to go and click connect more. All right. And what I just said to you, all the different connectors, they're all going to be in this browse connectors.

And this is also available if you go into settings. And then on the left hand side, go to connectors, but you can get there directly going the route that I just told you to. One thing to keep in mind, if you want to use model context protocol, that is Anthropics advanced kind of way that two different AI systems can talk to each other or AI systems like chat, GPT, Claude, co-pilot Gemini,

can talk to the rest of the internet, the way that websites have APIs, AI systems can talk to the rest of the web and other AIs using the MCP protocol or model context protocol. So if you want that, you can go into developer mode and toggle that on. However, if you do that to use custom MCPs, your normal connectors are not going to show up how they normally would. Just keep that in mind. All right. So I'm not going to go in and, you know, show you that. That's for another day. We'll do a dedicated MCP show. Again, I ask you guys, I never know, but let me know MCP yes or MCP no. Sometimes I say that and 50 people will reply.

Sometimes one person will reply. So let me know MCP yes or MCP no if we should do a dedicated episode on working with MCPs inside of chat. It's a little niche, but I also think it's extremely powerful, but also even opening I said it's extremely dangerous. Anyways, if you are looking for MCPs, that's where they are. They're not going to be under the browse connectors. All right. And then from there, what you would do, uh, let's say I want to connect my outlook email.

I don't really use outlook. Um, I use Gmail for the most part, but I would click Outlook email. All right. And then I would click, uh, this connect button right there. And then I would just continue to Outlook email. I'm not going to click that because then, uh, my email will be up there for everyone to see and I'm already bad enough at responding to, uh, your all. all very important emails.

All right. But then it's like there, like you would for any SSO connection, secure sign on, right, or single sign on. You just authorize it. You give it, you know, read, view access. And then you're ready to go, right? It is that simple. Like I said, again, this isn't as powerful and robust and as secure technically as, as working

with traditional rag. But you don't got to spend six or seven figures and you don't got to wait six to 12 months. get it, a couple clicks of the button. Again, make sure, check with whoever is in charge of your data and security on your team that you have access to do that. If you are on a teams or enterprise plan, you're not even going to be able to do this unless it's already allowed within an admin of your organization. So if you aren't a teams or enterprise plan, you're like, why can I do this?

Jordan, you're saying all these things. You're wrong. It's not there. It's because your admin hasn't enabled that. All right. But it's literally that simple. All right. So I showed you how to add a new connector.

All right. And then from there, you can select multiple connectors when you are doing a prompt. And the good thing is chat GPT obviously still retains all of its normal capabilities when you are putting in a prompt. So it still has all the kind of agenic scaffolding that a model like GPT5 or GPT5 thinking should have. Right? It can think, it can still plan ahead. it can go back and forth between different connectors.

It can, you know, start using information from some of your connected integrations. Then it can use advanced data analysis, start writing Python. It can then go, you know, jump and browse the web for something. Then it can go back to one of your other connectors. Again, it can't write, you know, to your files inside Google Drive, or it can't respond to emails just yet, although I do know that that will be coming at some point soon. But again, it can read, view, and share information across Connect. connectors, as well as it's still, you're still working with that big brain.

Right. I'm my model of choices, normally a GPT5 thinking or a GPT5 pro. Unfortunately, connectors don't work with pro. So in this case, I'm using GBT5 thinking. And again, keep in mind that you still have all the capabilities. I think a lot of times when I see people talking about connectors, they're just using it as like a one trick pony. And I'm like, well, why are you only doing it like that? You should be, you know, having chat, GPT pull and collaborate with the information between multiple connectors and then go help you get some work done.

All right. So I did start right before I started. I started a prompt because I figure it's going to take a couple of minutes. So let's go ahead and check in on that. Oh, cool. It's actually done. So here's what I said. All right.

So this, let's see how long this took. It took about four minutes and 53 seconds. So here's what I said. I said, and I intentionally try to throw chat chivity off and we'll see if it took the bait or if it did things correctly. I said, please check my Google Calendar and Gmail connected here. So again, I'm going to this chat and I can show you what's connected. My Gmail is auto.

That connector is searched automatically because I already have it connected. My Google calendar is auto. And then I had Google or sorry, I had Canva. selected on Google Drive selected on. So again, you can toggle these connectors on or off. You don't just like, let's say you connect 10 of them. You don't want to have all 10 of those toggles toggled on, especially if you are giving it a complex prompt.

Because chat, GPT, again, it's generative. And it might decide, oh, I need to look in your CRM for some of this information, right? You're asking me about an email and your calendar. I should just double check your CRM. And then all of a sudden, you're waiting way too long. So make sure don't just keep all of those, all of those integrations toggled on because sometimes you should still call out and say, go use this connector, but sometimes you don't have to and it's smart enough to know.

But there's a flip side to that. You might be waiting way longer than you may think or you may want if you leave too many of those connectors toggled on. All right. So let's quickly get to this use case, my example. And this is one way that I'm using connectors all the time. So I said, please check my Google calendar and Gmail connected here. I have a podcast recording next week with WWT.

I forgot who it's with. Research them. There might be duplicates on my calendar, so check it all. Tell me the details from that invite, then find the corresponding email in my Gmail. After that, please carefully look through my Canva account and find relevant docs to the show discussed. Essentially, I'm going on, I had WWT, they're a huge tech company, one of the biggest tech companies in the world. I had their CEO on my podcast like two years ago. I'm going on their fantastic podcast here pretty soon.

And this is what I normally, I waste so much time, right? Because I have to open my calendar. I have to open my email. Usually there's multiple threads within my email. Then I'm having to do some research, right? If it's like, oh, yeah, yeah, yeah. I'm going to talk about, you know, I don't know, the agent washing, right? Oh, that was in my Canva document, right?

I, a lot of my podcast, well, every single podcast, I always have slides and I put some notes in there, right? It's still technically unscripted. I'm still riffing off the top. But I have a lot of good information that just lives in my Canva. And it's hard to pull things out of there, right? I literally have 600 essentially PDFs that I've created with a ton of great information. So what I would normally do before connectors and before kind of chat, EBT is I would have to open multiple email threads. I'm going to get distracted.

I'm going to see an important email that I forgot to reply to. I'm going to have to go to reply to that. Then I'm going to open my calendar. Oh, crap. I have this, you know, I have this calendar invite that I got to go check, right? It's so easy to get distracted, especially if you're a small business owner like me and you're wearing 82 different hats. It is so difficult, right?

And then I would have to go, you know, do a little bit of research on the show, pull some of my information, start compiling it, all this, right? To normally, before AI and before chat, GPT, before connectors, this little project here would take me three to six hours, right? It would take me a long time. Go through, read all these emails, take notes, go do some research, go find my old information of things that they want to talk about. They told me some things they want to talk about. I know I've covered it all. I got to go find all the information. I got to start copying and pacing all this, right? Nope, not anymore. I use connectors. All right. Let me keep going. So I said, then find the corresponding email in my Gmail.

After that, please look carefully through my Canva account and find relevant docs to the show discussed. There's a ton of Canva docs, so please go step by step and search deeply. I said, in your reply back, do not include email addresses. I didn't want anyone's email address from WWT to be exposed here on the live stream. And then I said, just use their names and other details to accommodate. And then I have something in there about my custom instructions to get going through a little more quickly. And then I said, in this task, please reply back with specific bullet points showing the info

from my calendar, the email recap based on that calendar event. and then relevant info from my Canva documents. And I'm saying, give me actual specifics, not just like, hey, in this Canva document, you talk about blah, blah, blah. And then I said also additionally research the web about topics not covered in my Canva Docs, but that are discussed in that email. And then I just said reply back with easy to read formatting. And for whatever reason, GPD5 thinking just doesn't, can't format.

So I'm actually going to quickly switch over to GP5 auto to, essentially just reformat this. But let's look through and see what happened. Okay. So now I'm going to click on the thought and I can kind of show you what's happening. Again, a lot of this is trying to avoid some of my custom instructions and it's overriding them. All right, but essentially what's happening here is it's going through my calendar and it finds out and here's the trick. I said next week, it's actually not next week. It's tomorrow. All right. And it looks like chat GPT and the connectors was smart enough. It found the WD. WT references in my email and on my calendar.

And it's like, yo, Jordan, it's not next week. It's actually tomorrow. So it went through and it found it searched in my calendar for anything. WWT. It found it.

It looks like that is happening tomorrow. There we go. It's the people who it's with. And then it's searching for that calendar event and some of those keywords, the people I'm meeting with in my Gmail. All right. Then it found 10 different emails. We weren't collaborating on all of those tens.

I've just emailed some of these people before dating back a couple of years ago. Right. So it's going through. It's looking at all of those emails to surface just the relevant information that aligns with the calendar invite. All right. And then from there, it says, I'll now search the Canva documents with multiple queries, including the terms. Right. So it's searching terms like WWT Podcast, Worldwide Technology, AIPG Podcasts, right?

some of the keywords that it found in the, in the email. And then it's going through. It's doing some, some more, some more searches. It's searching now, after all that,

it went through all my documents. Now it's finding out information that I didn't have in my Canva docs to go find, right? I told it, hey, who is this podcast with, right? I know who it's with.

But I, well, I actually, I haven't met, everyone that's going to be on the call. So it is going to do some more additional research for me. All right. So it's searching on their website. It's finding information about these people that I'm meeting with. It looks like it's going in there and looking at their podcast. It's great podcast. By the way, give it a plug. It's called WWT's AI Proving Ground. All right. So it's looking at

that podcast, who they partner with, what they normally cover, the topics that we discussed, right? You see all of this information. It is going to, out synthesizing information from all of my business context. Then it is going through. It is researching based on my business context and what I told it to do. It's synthesizing all this information and then personalizing it exactly how I need it. All right. So then I can go down and bam.

There we go. All right. So here it is the calendar info. It found the correct time. It gave me the details from my calendar. Then here's the email recap based on the calendar. There we go. some of the things that we're going to be covering.

We're going to be, they want to talk about my roasting of the MIT study. Can't wait. It's going to be a fun one. So make sure you tune into the WWT AI Proving Ground podcast. All right. There we go. And then it's bringing in relevant info from my Canva documents.

Perfect. And again, this is where citing in sources matter. So it's pulling in and you'll see this, the Canva integration is actually really good. Because I do a bad job of naming my file. This one is called copy of copy of copy of copy of copy of copy of copilot free. Right?

I should probably start naming them a little bit better. But that just shows you how good of a job. And I wasn't even in the deep research mode. I was technically in the chat mode, but I was using GPT5 thinking. So it did a really good job. You know, it's a model with agentic capabilities. It did a good job of digging deep as I told it to. And then it finds, even though I'm not naming my files correctly,

it's finding context inside of those Canva documents, right? And then I can click on these, right, if I want to double check something. So it looks like from Microsoft build, all right? So it looks like something in there is related to what we may be covering on the podcast, but I don't even have to go in and read it if I don't want to because it broke down everything that we talked about, right? So it looks like we'll probably be talking about multi-agentic orchestration and co-pilot the studio, A to A protocol, MCP protocol, support across GitHub, Azure Dynamics, Windows 11,

right? So it's all right there, right? I spent so much time putting all this information together over the years. So I don't have to spend three to six hours going to prepare for it. It's all right here. Yeah, I did mention GP5 thinking not great at, not great at, you know, having nicely formatted information there. But at the bottom, all of this information is cited at the bottom. So I can go click and check everything. Fantastic. And then I just ran another prompt that just said, you know, format this better.

And then I used GPG5 Auto. And there we go. So that's one quick example. But you'll see what I did there. I'm using. In that instance, I use three different connectors. All right. Because again, if you really want to get the true,

utility and the true time savings, the true ROI on this, don't, don't think of it as just like uploading a file, right? Think of those instances like, like myself, maybe where you personally struggle to focus, to concentrate things that you just find difficult, right? Maybe your share point, you just stink finding information across there, but chat chbt is great. Or when you go into HubSpot, right, you get lost inside looking for things. I do, right? the chat gvety connector is fantastic for hubspot right and you can go and say hey check you know check my hub stop check check my emails you know check my team's messages and tell me what i should be focusing on today that right there you can run that every single day and it's probably how you might

be right let's just say you know you use hub spot and teams and uh gmail right that could be how you spend most of your first two hours of the day and you can get a good kickstart and 10 or 15 minutes and keep refining and improving your prompt day by day. Again, think of balancing between multiple connectors and then having a very smart assistant, an assistant that's smarter than you in GPT5. Just go ahead and use GVT5 thinking for me. It's worth the extra, you know, four minutes there. A lot of times I'll just run multiple these prompts.

Go get my coffee upstairs, come downstairs, and I'm ready to go. That's how you need to be thinking of it. All right. So there we go. We have a demo that didn't break. break. All right. Live stream audience. What do you think? Impressive? Yon. Are you ready to put AI to work on this Wednesday? All right. Now let's very quickly go over some differences between rag and connectors and then talk a little bit about some use cases.

We're going to go quick here. So buckle up y'all. Buckle up buttercup. Here we go. So what's the difference? Is this rag? Not really. Can it. substitute for rag, traditional large language model rag in some use cases, sure. All right. So both retrieval. So the similarities between connectors and traditional rag. Well, they both retrieve external data first, then they generate answers, minimizing hallucinations.

That's number one. So traditional rag obviously requires custom pipelines and beddings and some database engineering. Connectors plug and play. Hardly no setup. couple clicks, you're ready to go. And they're maintained and updated automatically by Open AI. Some of the big differences.

Well, you saw connectors, depending on if you're on a team plan, enterprise plan, pro plan teams, right? But I mean, you're looking at anywhere from $20 to $200 a month per user, actually probably $100 or yeah, $200 a month for the pro plan per user. And that's instant deployment, no technical overhead. And then with traditional. very different, right? Yeah, it's getting much easier to have actual, you know, rag instances out the door working with different large language models. It's much easier

than it used to be, but still, traditional rag projects still might cost anywhere from 10,000 to 500,000 or more, but obviously higher accuracy and lower per query cost at scale, right? But the best fit, I think connectors for most businesses will work, right? And I think RAG for highly specialized or high value or high volume use cases, right? If you're a Fortune 100 company and whatever that you're trying to automate is the backbone of your business, you're probably going to be way better off with RAG, especially because right now connectors don't have read right. You know, default connectors just have read access, right? But Ragn can be a little different.

Now let's talk a little bit about MCP and how you can actually really expand capabilities with custom connectors via MCP. So Anthropics model context protocol, let developers build connectors for proprietary systems. So right now, you know, I rattled off about a dozen, 15 different connectors that are supported and created by Open AI. But you can literally do this with anything. All right. And MCP is essentially a USBC for AI.

It lets AI talk to other AI and different large language models talk to essentially a bunch of internet websites. And this does also support future right actions. So updating CRMs, ticketing tools or calendars with approval. All right. Let's go over three, what I think are pretty cool use cases for chat chipped deconector. So number one, preparing for a high stakes client meeting. I kind of give you an example, a meeting I'm prepping for. It's not high stakes.

It's going to be a fun conversation. But think what are those? And when you think about use cases, I rattled off all those connectors, write down the ones that you use, and then start tracking. How are you wasting your time in there? Go in your internet browsing history and see, oh my gosh, I spent four hours today in HubSpot. Oh my gosh.

I spent, you know, two and a half hours in, you know, teams and SharePoint today. Why? Right? You have to almost manually audit yourself to start building these use cases. So use case one, preparing for a high-stakes client meeting. So without connectors, you're going to have to manually search HubSpot for client history, Gmail for recent emails, and then Google Calendar for upcoming meetings. That could easily take you 30 minutes to a couple of hours.

So with connectors, enable HubSpot, Gmail, Google Calendar, and then ask, summarize this client status using CRM history, latest emails, and upcoming meetings, and then go ahead and have it do some web research for you as well. And then the outcome, well, consolidated, verifiable briefing across three different systems, probably saving more than 90% of the time if you were doing it manually without AI slash chat EBT. Use case two. Answering HR or compliance policy questions.

Don't we all love my HR people, don't you just love answering the same questions over and over and over? What if you could empower people to just get those answers themselves? All right. So without connectors, you're going to have to dig through your SharePoint HR folders, open PDF policies, read them, summarize them, and maybe ask HR staff. You take a lot. One simple query could easily take 15 minutes, an hour, maybe more.

With connectors, go ahead, connect SharePoint, One Drive, and Outlook, and ask, what is our PTO carryover policy and who approves exceptions? 30 seconds, you're going to get instant, cited, and answered, pull from synced HR handbooks, flagged Outlook memos, and official documents eliminated, wasted time searching. All right, use case number three, generating cross-functional reports. Here's a great one. Without connectors, this is a head. So you got to export Excel spreadsheets, compare data in Google Drive, then manually draft a summary and Word.

So with connectors, you can connect Google Drive, One Drive, and SharePoint, and ask. Summarize quarter three budget variance across finance spreadsheets and draft an executive summary. A couple of minutes versus a couple of hours. Then you have a reliable, grounded report combining spreadsheet and documentation from multiple repositories, cutting reporting time by probably more than 90%. all right so let's wrap up here with the question of why why should business leaders be using chat chbt connectors now well if you're not already convinced from what i already went over that's fine i'm not here to win you over open a i's not paying me a dime to say any of this but i want you to get more

out of a i think chat chvdd connectors again as long as you have access to do it it is a a no-brainer. You need to rely on them, not just use them, rely on them. Because not only is it instantly going to save you more time trying to find the files that you may be updating manually, you know, going through a, you know, kind of proper quote unquote context engineering process, but you're cutting down hallucinations and you're increasing the accuracy in richness of every single response when you use connectors. So you're going to get measurable productivity gains, hours saved weekly across sales, HR, finance, and ops. And connectors, unify silo data.

That's the biggest thing into a single reliable source of truth in chat GBT. That's the thing. Your data does not have to live in silos, right? If you use, you know, HubSpot, Gmail, and Notion, right? Those systems by default can't talk to each other. ChatGPT can make them talk to each other. and then you can create new business value out of that, out of those relationships. And then like I said, for most organizations, connectors deliver, I don't know,

80% of the benefits of traditional retrievalogue management generation at 2% of the cause and 1% of the time required. So again, this is not going to replace, right, if you're a Fortune 100 company, this is not going to replace an intricate and robust rag setup. but in many instances, you might not need it. Right. Like I said, I think you get about 80% of Rags benefits at 2% of the cost and 1% of the time required. And if that's not putting AI to work for you on Wednesdays, I don't know what it is.

All right. I hope this was helpful. We went over how connectors work. They are a secure bridge, right, between Chad GBT's default responses, which can absolutely stink and be riddled with hallucinations. They bridge your data. So you're not going to be getting these generic and maybe incorrect answers as frequently when you are using connectors and using them in the right way to share context and to create

a smaller, more specific version of chat GPT based on your company's data. And like I said, you shouldn't just be using them. You should be relying on them. If this episode was helpful, please find the LinkedIn. Maybe you're watching this on LinkedIn. maybe you're listening on the podcast. Go ahead. Check the show notes.

If you're listening on the podcast, we're going to have it on our newsletter everywhere else. Now, go repost this LinkedIn live stream. And I'm going to send you the chat GPT connectors cheat sheet. Because here we are. Obviously, I went 44 minutes. But I could have went for another three hours because there are so many great use cases, so many great tips and tricks on how to properly use connectors,

especially stacking these connectors with the power of GPT. and especially GPD5 thinking. So if you want the cheat sheet, I'm telling you, I put a ton of work on this one. You're going to want it. So make sure you go repost this episode. Sometimes I'll send it to you within an hour or so. Might take me a couple of days depending on when you repost it.

Just go ahead and repost it. Hit me with a message if you don't get it right away and I'll send it to you. Trust me, you're going to get a ton of value from that. So if you haven't already, please go to your everyday AI.com. Sign up for the free daily newsletter. Thank you for tuning in. Hope to see you back tomorrow and everyday. 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. dot adobe.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.