WEBVTT 00:00:00.000 --> 00:00:02.100 Have you ever found yourself kind of caught in 00:00:02.100 --> 00:00:05.679 that loop with ChatGPT? You know, where you're 00:00:05.679 --> 00:00:07.940 starting a brand new conversation every single 00:00:07.940 --> 00:00:10.460 time. And all that valuable context you just 00:00:10.460 --> 00:00:14.199 built up just sort of vanishes. Oh, tell me about 00:00:14.199 --> 00:00:16.410 it. It's a bit like Groundhog Day for your digital 00:00:16.410 --> 00:00:18.609 workflow, isn't it? Absolutely. You're constantly 00:00:18.609 --> 00:00:21.449 re -explaining the background, re -uploading 00:00:21.449 --> 00:00:24.329 the same files, repeating preferences over and 00:00:24.329 --> 00:00:26.890 over again. It's a genuine productivity drain. 00:00:27.050 --> 00:00:29.230 It really is. It's frustrating. But what if there 00:00:29.230 --> 00:00:32.689 was a smarter way, a way to actually organize 00:00:32.689 --> 00:00:36.630 your AI work and consistently get really high 00:00:36.630 --> 00:00:39.869 -quality results back? Well, today, we're taking 00:00:39.869 --> 00:00:42.270 a deep dive into exactly that. We're going to 00:00:42.270 --> 00:00:45.369 unpack ChatGPT projects. We'll look at what they 00:00:45.369 --> 00:00:47.750 are, why they're actually pretty powerful, and 00:00:47.750 --> 00:00:50.329 how you can set them up to hopefully transform 00:00:50.329 --> 00:00:52.630 your productivity. Yeah, let's do it. We can 00:00:52.630 --> 00:00:54.770 walk through some real examples, too, step by 00:00:54.770 --> 00:00:58.130 step. Perfect. So let's unpack this. So thinking 00:00:58.130 --> 00:01:00.469 back, most of us, when we first started with 00:01:00.469 --> 00:01:02.750 ChatGPT, we probably used it like a simple Q 00:01:02.750 --> 00:01:05.250 &A tool, right? Yeah, basically. You type a question, 00:01:05.370 --> 00:01:07.890 you get an answer, maybe a follow -up. Exactly. 00:01:08.090 --> 00:01:10.989 But the frustration really kicks in when you 00:01:10.989 --> 00:01:13.430 move to the next task, even if it's closely related. 00:01:14.269 --> 00:01:16.549 You feel like you have to start all over. You 00:01:16.549 --> 00:01:19.390 do. You hit that reset button. You find yourself 00:01:19.390 --> 00:01:21.709 re -explaining the whole context, your business, 00:01:21.909 --> 00:01:23.629 the project, the audience you're talking to. 00:01:23.730 --> 00:01:26.010 Re -uploading documents. Re -uploading, yeah. 00:01:26.530 --> 00:01:28.810 Reiterating the formatting you want. And the 00:01:28.810 --> 00:01:31.890 AI just starts completely fresh. Zero memory. 00:01:32.079 --> 00:01:34.239 of what you just did. That starting from scratch 00:01:34.239 --> 00:01:36.280 problem, it's such a real bottleneck. We saw 00:01:36.280 --> 00:01:38.299 that a lot in the user feedback people shared. 00:01:38.400 --> 00:01:41.379 It wastes so much time and maybe even more importantly, 00:01:41.439 --> 00:01:44.379 it leads to really inconsistent results. Right. 00:01:44.700 --> 00:01:47.359 I mean, think about it. If the AI doesn't remember 00:01:47.359 --> 00:01:51.219 your brand specific tone or your clients unique 00:01:51.219 --> 00:01:53.799 needs, how can it possibly give you the right 00:01:53.799 --> 00:01:56.459 output every time? Exactly. It can't. And this 00:01:56.459 --> 00:01:59.040 is precisely the problem that chat GPT projects 00:01:59.040 --> 00:02:01.359 are trying to solve. They're not just, you know, 00:02:01.359 --> 00:02:04.000 glorified folders for storage. That's a common 00:02:04.000 --> 00:02:06.200 misconception, isn't it? Yeah, it really is. 00:02:06.319 --> 00:02:10.159 Instead think of them more like dedicated intelligent 00:02:10.159 --> 00:02:13.280 workspaces. Okay. Digital environments built 00:02:13.280 --> 00:02:16.240 specifically for that ongoing multi -step work 00:02:16.240 --> 00:02:20.180 where context is absolutely critical. So whether 00:02:20.180 --> 00:02:22.919 you're managing say content creation for a client 00:02:22.919 --> 00:02:25.719 or a big marketing campaign or even writing a 00:02:25.719 --> 00:02:29.039 book. Yeah, anything complex or ongoing. Projects 00:02:29.039 --> 00:02:32.460 aimed to keep everything organized and crucially, 00:02:32.460 --> 00:02:34.680 contextually aware. That's the key. Every new 00:02:34.680 --> 00:02:36.639 chat you start inside a project, it's already 00:02:36.639 --> 00:02:39.360 preloaded. How so? It gets immediate access to 00:02:39.360 --> 00:02:41.680 all the relevant past conversations within that 00:02:41.680 --> 00:02:43.900 project, all the files you've uploaded, like 00:02:43.900 --> 00:02:46.740 brand guidelines, research docs, customer data, 00:02:46.860 --> 00:02:50.560 and maybe most importantly, your custom instructions, 00:02:50.800 --> 00:02:53.560 which means you literally never start from zero 00:02:53.560 --> 00:02:57.060 again. every conversation builds on that foundation. 00:02:57.300 --> 00:02:59.740 Leading to better, more consistent results, presumably. 00:02:59.800 --> 00:03:02.419 Way more consistent, yeah. And just a quick practical 00:03:02.419 --> 00:03:04.419 note here, based on the articles we looked at, 00:03:04.560 --> 00:03:06.939 projects are currently only for chat GPT plus 00:03:06.939 --> 00:03:09.280 and pro users. Right, so if you don't see projects 00:03:09.280 --> 00:03:11.419 in your sidebar, that's probably why. Probably. 00:03:11.740 --> 00:03:14.379 So, okay, beyond just saving a few clicks here 00:03:14.379 --> 00:03:17.860 and there, what's the real, like, core difference 00:03:17.860 --> 00:03:21.360 this brings to how we use AI day to day? It's 00:03:21.360 --> 00:03:23.379 an intelligent workspace, no more starting over. 00:03:23.680 --> 00:03:26.550 Got it. Let's talk practical setup then, based 00:03:26.550 --> 00:03:30.370 on the guide. Step one seems simple enough. Just 00:03:30.370 --> 00:03:32.349 creating the project. Yeah, pretty straightforward. 00:03:32.930 --> 00:03:35.669 On the left side of your chat GPT screen, you'll 00:03:35.669 --> 00:03:38.229 see a new project button, usually with a little 00:03:38.229 --> 00:03:40.710 plus icon. Click that and it kicks things off. 00:03:41.169 --> 00:03:44.669 Yeah. Now, this next bit seems surprisingly important, 00:03:44.969 --> 00:03:48.030 but maybe easy to overlook, naming it. Oh, definitely. 00:03:48.349 --> 00:03:50.669 Give it a clear descriptive name. Don't just 00:03:50.669 --> 00:03:52.840 call it project one. Right. So like the example, 00:03:52.919 --> 00:03:54.740 if you're Momentum Digital working for TechUp, 00:03:54.939 --> 00:03:57.080 maybe Momentum Digital Marketing for Client TechUp. 00:03:57.360 --> 00:03:59.080 Exactly. Something where you can just glance 00:03:59.080 --> 00:04:00.860 at it and know exactly what's inside. Then just 00:04:00.860 --> 00:04:03.620 click Create Project. OK. Project created. It 00:04:03.620 --> 00:04:05.400 pops up in the sidebar. And then you mentioned 00:04:05.400 --> 00:04:08.000 a visual tip. Yeah. A simple but helpful one. 00:04:08.060 --> 00:04:10.280 You can change the folder color. Just click the 00:04:10.280 --> 00:04:12.199 little folder icon at the top of the project 00:04:12.199 --> 00:04:15.020 page. And you can color code things. Maybe blue 00:04:15.020 --> 00:04:18.259 for client work. green for internal stuff, red 00:04:18.259 --> 00:04:20.920 for urgent tasks, whatever system works for you. 00:04:21.060 --> 00:04:23.620 Nice visual cue. And then step three is choosing 00:04:23.620 --> 00:04:26.420 the AI model. Right. Projects let you use different 00:04:26.420 --> 00:04:28.639 models within the same workspace. You've got 00:04:28.639 --> 00:04:32.720 GPT -4 .0 for the complex, creative stuff, GPT 00:04:32.720 --> 00:04:35.180 -4 .0 mini for maybe quicker, simpler things, 00:04:35.699 --> 00:04:38.759 and GPT -3 .5 for really basic stuff. But for 00:04:38.759 --> 00:04:40.860 most serious business tasks, like marketing. 00:04:41.079 --> 00:04:43.800 Yeah, for that nuanced understanding and quality, 00:04:43.980 --> 00:04:46.220 GPT -4 was probably still your best bet. That's 00:04:46.220 --> 00:04:48.139 what we generally recommend sticking with for 00:04:48.139 --> 00:04:50.379 the core work in a project. OK, makes sense. 00:04:50.759 --> 00:04:53.100 So how does this initial setup, these first few 00:04:53.100 --> 00:04:55.819 steps, really benefit our overall workflow in 00:04:55.819 --> 00:04:58.759 the long run? Visual cues and model choice streamline 00:04:58.759 --> 00:05:01.279 your process. OK, now we get to what the guides 00:05:01.279 --> 00:05:05.540 and articles. really highlight as the core components. 00:05:06.000 --> 00:05:08.199 The real magic, as some put it. Yeah, this is 00:05:08.199 --> 00:05:10.699 what turns it from just a folder into that intelligent 00:05:10.699 --> 00:05:12.860 workspace we talk about. And that's files and 00:05:12.860 --> 00:05:15.399 custom instructions. Exactly. The two pillars. 00:05:15.740 --> 00:05:18.540 Let's start with files. Uploading relevant files. 00:05:18.879 --> 00:05:20.459 You said these become like a dedicated knowledge 00:05:20.459 --> 00:05:23.060 base. Precisely. It's like giving ChatGPT its 00:05:23.060 --> 00:05:26.160 own private library just for this project. Only 00:05:26.160 --> 00:05:28.720 the relevant info. So for that marketing project 00:05:28.720 --> 00:05:30.920 example, what kind of files are we talking about? 00:05:31.000 --> 00:05:33.879 OK, so for our momentum digital tech up example, 00:05:34.199 --> 00:05:36.600 you definitely want client documents, things 00:05:36.600 --> 00:05:39.579 like detailed target audience personas, user 00:05:39.579 --> 00:05:43.139 pain point analysis, market research, competitor 00:05:43.139 --> 00:05:45.740 analysis. Absolutely. Then you layer on your 00:05:45.740 --> 00:05:48.160 brand materials, the detailed guidelines, tone 00:05:48.160 --> 00:05:51.240 of voice docs, maybe even examples of past successful 00:05:51.240 --> 00:05:53.819 campaigns. Got it. And don't forget product or 00:05:53.819 --> 00:05:56.939 service info, feature descriptions, pricing, 00:05:57.620 --> 00:05:59.860 technical docs maybe so the AI really gets what 00:05:59.860 --> 00:06:02.509 you're selling. And social proof, testimonials, 00:06:02.709 --> 00:06:05.250 case studies. Super valuable, yeah. Ads credibility. 00:06:05.509 --> 00:06:07.370 And uploading is easy. Just click add files, 00:06:07.490 --> 00:06:09.430 grab your documents. You can do multiple at once. 00:06:09.889 --> 00:06:11.750 Build up that knowledge base. OK, that's the 00:06:11.750 --> 00:06:15.490 files foundation. Now, custom instructions. You 00:06:15.490 --> 00:06:17.189 said this is where the magic really happens. 00:06:17.529 --> 00:06:19.889 Yeah, I think so. This is like giving ChatGPT 00:06:19.889 --> 00:06:22.509 its detailed brief, almost its personality for 00:06:22.509 --> 00:06:24.810 this specific project. So what makes for great 00:06:24.810 --> 00:06:27.350 custom instructions? It can't just be, be helpful. 00:06:27.629 --> 00:06:30.769 No, definitely not. You need specifics. First, 00:06:31.170 --> 00:06:33.829 assign it a clear role. Tell it who it should 00:06:33.829 --> 00:06:37.639 be. expert consultant, a creative copywriter. 00:06:37.680 --> 00:06:41.740 OK. Then specify the response format you want. 00:06:42.240 --> 00:06:45.000 Define the tone and style really clearly. Include 00:06:45.000 --> 00:06:47.439 any unique guidelines just for this project. 00:06:47.660 --> 00:06:49.839 Reference points, too, like specific authors 00:06:49.839 --> 00:06:53.139 or styles to emulate. Yeah, exactly. So for our 00:06:53.139 --> 00:06:55.300 TechUp B2B example, the instructions might be 00:06:55.300 --> 00:06:57.920 something like, core role, you are a growth marketing 00:06:57.920 --> 00:07:01.160 consultant for B2B tech startups. OK, very specific. 00:07:01.339 --> 00:07:03.920 Focusing on attracting technical audiences for 00:07:03.920 --> 00:07:06.269 style. Inspired by Seth Godin, professional, 00:07:06.529 --> 00:07:08.910 tech savvy, focused on benefits, not just features. 00:07:09.029 --> 00:07:11.170 Nice format. Structured recommendations, use 00:07:11.170 --> 00:07:13.790 bullet points, headings, always suggest a title 00:07:13.790 --> 00:07:15.949 and a call to action for content. And the business 00:07:15.949 --> 00:07:19.350 context. Client is tech up, saws for developers. 00:07:20.029 --> 00:07:24.329 Target, devs, tech leads, CTOs, goal, increase 00:07:24.329 --> 00:07:27.389 trial signups, avoid cliche marketing jargon, 00:07:27.629 --> 00:07:30.069 be authentic. That's pretty detailed. It needs 00:07:30.069 --> 00:07:32.420 to be. You know, I still wrestle with prompt 00:07:32.420 --> 00:07:34.579 drift myself sometimes. Yeah, where it kind of 00:07:34.579 --> 00:07:37.220 loses focus. Exactly. It starts to wander off 00:07:37.220 --> 00:07:39.439 from the original instructions over a long chat. 00:07:40.079 --> 00:07:42.360 Getting these custom instructions really dialed 00:07:42.360 --> 00:07:46.360 in is, it's an art, beat, but when you nail it, 00:07:46.680 --> 00:07:49.160 wow, it makes such a difference. It really sounds 00:07:49.160 --> 00:07:51.379 like these two things, the file providing the 00:07:51.379 --> 00:07:53.480 knowledge, the instructions providing the personality 00:07:53.480 --> 00:07:56.089 and direction are the absolute heart of it. Why 00:07:56.089 --> 00:07:58.050 are they so fundamentally critical together? 00:07:58.189 --> 00:08:00.930 They provide the core brain and personality for 00:08:00.930 --> 00:08:03.689 your AI. Okay, let's make this concrete. A real 00:08:03.689 --> 00:08:05.750 world use case, like the ones in the studies, 00:08:06.230 --> 00:08:08.750 say creating an outline and intro for a technical 00:08:08.750 --> 00:08:11.370 blog post. Alright. For our TechUp example, maybe 00:08:11.370 --> 00:08:15.290 the task is detailed outline and intro for automating 00:08:15.290 --> 00:08:18.129 your API testing workflow with TechUp's new feature. 00:08:18.410 --> 00:08:20.509 So instead of just that generic prompt, you'd 00:08:20.509 --> 00:08:23.439 leverage the project context. Right. Specify 00:08:23.439 --> 00:08:26.399 the target devs, QA engineers. Address their 00:08:26.399 --> 00:08:29.500 specific pain points with manual testing. Introduce 00:08:29.500 --> 00:08:32.019 the new feature as the solution. Right. Build 00:08:32.019 --> 00:08:34.620 credibility. Maintain that tech savvy tone from 00:08:34.620 --> 00:08:37.460 the instructions. And maybe specify the format 00:08:37.460 --> 00:08:41.840 detailed outline with HUSH3s and like a 150 -200 00:08:41.840 --> 00:08:45.220 word. And with a well set up project, the output 00:08:45.220 --> 00:08:47.419 should be pretty impressive. You'd get that detailed 00:08:47.419 --> 00:08:50.480 outline, H2s like the problem with manual API 00:08:50.480 --> 00:08:53.059 testing, H3s like time drain, scaling issues, 00:08:53.220 --> 00:08:55.519 and then a solid intro. Something like, if you're 00:08:55.519 --> 00:08:58.759 a developer, you know the pain. Hours on repetitive 00:08:58.759 --> 00:09:01.360 API tests. It's a bottleneck. What if you could 00:09:01.360 --> 00:09:03.429 automate it? TechUp's new auto -test feature 00:09:03.429 --> 00:09:04.970 makes that real, you know, something engaging 00:09:04.970 --> 00:09:06.929 like that. And it works well because the AI has 00:09:06.929 --> 00:09:09.129 all that context from the start. Exactly. It 00:09:09.129 --> 00:09:11.210 knows the audience, the product, the desired 00:09:11.210 --> 00:09:13.889 tone from the files and instructions, so it maintains 00:09:13.889 --> 00:09:17.049 consistency, delivers professional quality, it 00:09:17.049 --> 00:09:19.289 has guardrails. That's pretty efficient. OK, 00:09:19.289 --> 00:09:21.129 now let's take that blog post and promote it. 00:09:21.129 --> 00:09:23.830 OK. Use case hashtag two, a LinkedIn post series. 00:09:24.169 --> 00:09:27.610 Good one. Now, this is a related task, but different 00:09:27.610 --> 00:09:30.690 social media, not long -form content. So key 00:09:30.690 --> 00:09:34.399 step. Start a new chat within the same project, 00:09:34.480 --> 00:09:37.139 right? Crucial step. Yep. Keeps things organized. 00:09:37.379 --> 00:09:39.840 But you still get all that lovely context. Tech 00:09:39.840 --> 00:09:42.539 up, the product, the audience, the tone. It's 00:09:42.539 --> 00:09:44.960 all still there. Okay, so the prompt for the 00:09:44.960 --> 00:09:47.879 LinkedIn series might detail, say, a three -post 00:09:47.879 --> 00:09:50.659 strategy. Yeah, like, post one is a hook about 00:09:50.659 --> 00:09:53.690 the pain points. Post 2 introduces the blog feature 00:09:53.690 --> 00:09:57.029 with benefits and a CTA. Post 3 offers social 00:09:57.029 --> 00:09:59.470 proof or a future vision. Makes sense. And it 00:09:59.470 --> 00:10:01.710 would generate those posts. Yep. Like Post 1 00:10:01.710 --> 00:10:04.470 might be. Still manually testing APIs, flaky 00:10:04.470 --> 00:10:07.009 tests, regressions. It's a productivity killer. 00:10:07.409 --> 00:10:09.210 There is a better way. TechUp's working on something. 00:10:09.309 --> 00:10:11.429 Stay tuned. You know, with tailored hashtags, 00:10:11.629 --> 00:10:13.820 too. And the power here is that consistency. 00:10:14.379 --> 00:10:16.039 Absolutely. The project continuity. The whole 00:10:16.039 --> 00:10:17.720 series sounds like it came from the same brand 00:10:17.720 --> 00:10:20.519 as the blog post. Unified voice, coherent message, 00:10:20.919 --> 00:10:22.840 totally brand aligned with TechUp's goals in 00:10:22.840 --> 00:10:25.379 the project. Wow. Think about scaling that unified 00:10:25.379 --> 00:10:28.059 voice. Whoa. Imagine scaling that across hundreds 00:10:28.059 --> 00:10:29.879 of campaigns without re -briefing every time. 00:10:30.159 --> 00:10:32.860 That's the real power. So it's not just faster 00:10:32.860 --> 00:10:35.519 content. How does this process really save us 00:10:35.519 --> 00:10:38.419 significant time and, frankly, mental energy? 00:10:38.590 --> 00:10:41.509 AI understands your brand, generating consistent 00:10:41.509 --> 00:10:44.149 tailored content faster and smarter. Okay, so 00:10:44.149 --> 00:10:46.490 to really get the most out of this, we need to 00:10:46.490 --> 00:10:49.470 avoid some common mistakes. Based on what we've 00:10:49.470 --> 00:10:52.429 seen in forums and guides, what are the big pitfalls? 00:10:52.750 --> 00:10:55.509 I think the biggest one is just treating projects 00:10:55.509 --> 00:10:58.350 like basic folders. You know, creating one, but 00:10:58.350 --> 00:11:00.809 then not actually using the files and custom 00:11:00.809 --> 00:11:02.769 instructions. Right. If you don't feed it that 00:11:02.769 --> 00:11:05.149 context, it's basically just an empty box. No 00:11:05.149 --> 00:11:07.409 different from a regular chat. Exactly. It defeats 00:11:07.409 --> 00:11:10.070 the whole purpose. Another one is uploading too 00:11:10.070 --> 00:11:12.710 many irrelevant files. Ah, information overload 00:11:12.710 --> 00:11:15.250 for the AI. Yeah, basically. If you just dump 00:11:15.250 --> 00:11:17.529 everything in there, it kind of dilutes the context. 00:11:17.730 --> 00:11:19.769 The AI struggles to figure out what's actually 00:11:19.769 --> 00:11:22.549 important for this specific task. So be selective. 00:11:23.149 --> 00:11:25.620 Curate the knowledge base. Definitely. Choose 00:11:25.620 --> 00:11:28.879 only the critical, relevant docs for that project's 00:11:28.879 --> 00:11:31.500 goal. And we absolutely have to talk about vague, 00:11:31.600 --> 00:11:33.659 custom instructions again. That seems like a 00:11:33.659 --> 00:11:36.779 major issue. Huge. Just saying, be professional 00:11:36.779 --> 00:11:39.840 or write well, that gives the AI almost nothing 00:11:39.840 --> 00:11:42.320 to work with. Useless. Pretty much. You need 00:11:42.320 --> 00:11:45.299 those hyper -specific details. The role, the 00:11:45.299 --> 00:11:48.120 audience, the tone, the business context, like 00:11:48.120 --> 00:11:51.480 we talked about, precision is key. Got it. Anything 00:11:51.480 --> 00:11:54.929 else? Yeah. One more big one. Trying to do too 00:11:54.929 --> 00:11:56.889 many different things in the same chat window? 00:11:57.190 --> 00:12:00.210 The kitchen sink approach. Exactly. Don't ask 00:12:00.210 --> 00:12:02.669 it to write a blog post, then immediately draft 00:12:02.669 --> 00:12:06.009 an email, then do social posts all in one continuous 00:12:06.009 --> 00:12:08.350 thread. Why not? Because the context gets modeled. 00:12:08.870 --> 00:12:11.590 Each task really benefits from a focused conversation. 00:12:11.899 --> 00:12:14.840 So start a new chat within the project for each 00:12:14.840 --> 00:12:17.159 distinct task, even if they're related. Keeps 00:12:17.159 --> 00:12:19.299 things clean and effective. Right. Otherwise, 00:12:19.559 --> 00:12:22.340 it's like trying to bake a cake, fix a car, and 00:12:22.340 --> 00:12:24.240 write a novel all in the same kitchen sink. It 00:12:24.240 --> 00:12:26.139 just gets messy and nothing turns out quite right. 00:12:26.399 --> 00:12:28.620 That kitchen sink analogy definitely sticks. 00:12:28.820 --> 00:12:30.980 Yeah. So if there's one big takeaway for listeners 00:12:30.980 --> 00:12:34.000 to avoid these traps, what is it? Be present 00:12:34.000 --> 00:12:36.639 with context and keep your AI tasks focused. 00:12:36.879 --> 00:12:39.500 Okay, one area of confusion that often pops up 00:12:39.500 --> 00:12:42.259 is the difference between chat GPT projects and 00:12:42.259 --> 00:12:44.600 custom GPTs. Let's try and clarify that. Yeah, 00:12:44.639 --> 00:12:46.879 good idea. They sound similar, but serve different 00:12:46.879 --> 00:12:50.320 purposes. Think of custom GPTs as specialized 00:12:50.320 --> 00:12:53.679 tools. They're built for specific, often repetitive 00:12:53.679 --> 00:12:56.039 tasks. They can have a unique personality, they're 00:12:56.039 --> 00:12:58.620 easily shareable, and they can even connect to 00:12:58.620 --> 00:13:01.899 external things like APIs. More like single purpose 00:13:01.899 --> 00:13:05.360 apps and projects. Projects are more about organizing 00:13:05.360 --> 00:13:08.019 ongoing evolving work. They're your personal 00:13:08.019 --> 00:13:11.120 workspaces. Their main job is storing files, 00:13:11.600 --> 00:13:14.299 remembering instructions, and keeping that rich 00:13:14.299 --> 00:13:17.100 chat history going over time for a broader initiative. 00:13:17.289 --> 00:13:20.210 So give us the when to use which breakdown. OK. 00:13:20.549 --> 00:13:22.990 Use projects for managing complex client work, 00:13:23.289 --> 00:13:25.529 multi -step content creation, needing a consistent 00:13:25.529 --> 00:13:28.090 voice, or any long -term thing where context 00:13:28.090 --> 00:13:31.590 builds up. And custom GPTs. Use custom GPTs when 00:13:31.590 --> 00:13:34.049 you want to build a specific repeatable tool, 00:13:34.250 --> 00:13:37.129 like an expert ad copywriter GPT that always 00:13:37.129 --> 00:13:39.750 uses a certain framework, or an AI assistant 00:13:39.750 --> 00:13:42.090 for your team that does one specific task often. 00:13:42.309 --> 00:13:44.289 They're for narrow specialized capabilities. 00:13:44.769 --> 00:13:47.639 Gotcha. But they aren't mutually exclusive not 00:13:47.639 --> 00:13:49.840 at all and this is where it gets really powerful. 00:13:49.899 --> 00:13:52.509 You can absolutely use them together How would 00:13:52.509 --> 00:13:55.190 that work? Well, our agency, Momentum Digital, 00:13:55.429 --> 00:13:57.490 could use a project for all the work for the 00:13:57.490 --> 00:14:00.210 TechUp client blogs, social strategy, everything. 00:14:00.950 --> 00:14:04.009 But they could also have a custom GPT, maybe 00:14:04.009 --> 00:14:06.789 called LinkedIn Post Optimizer, that the whole 00:14:06.789 --> 00:14:09.490 marketing team uses across all clients, including 00:14:09.490 --> 00:14:12.830 TechUp, just to refine those posts. Ah, so the 00:14:12.830 --> 00:14:16.690 Cumptive GPT is a tool used within or alongside 00:14:17.200 --> 00:14:20.059 the broader project structure. Exactly. It lets 00:14:20.059 --> 00:14:22.500 you layer workflows in a really advanced way. 00:14:22.639 --> 00:14:24.720 So these two powerful features, projects and 00:14:24.720 --> 00:14:27.059 custom GPTs, they can actually work together, 00:14:27.340 --> 00:14:29.240 complementing each other. Absolutely. They complement 00:14:29.240 --> 00:14:31.539 each other perfectly for advanced workflows. 00:14:31.960 --> 00:14:34.220 OK, let's wrap this up. What does this all really 00:14:34.220 --> 00:14:36.440 mean for your day -to -day workflow with AI? 00:14:36.960 --> 00:14:38.980 What are the big ideas? Well, first, I'd say 00:14:38.980 --> 00:14:42.379 setup is everything. Really take that time upfront 00:14:42.379 --> 00:14:44.940 to configure your project with detailed instructions 00:14:44.940 --> 00:14:47.360 and truly relevant files. That's the foundation, 00:14:47.460 --> 00:14:49.879 isn't it? It absolutely is. And that leads to 00:14:49.879 --> 00:14:53.580 the second point. Context is king. The more relevant, 00:14:53.740 --> 00:14:56.100 thoughtful context you provide, the better and 00:14:56.100 --> 00:14:58.279 more precise your results will be. The AI just 00:14:58.279 --> 00:15:00.919 gets smarter about your needs. And third? Organization 00:15:00.919 --> 00:15:03.980 pays off. A good project structure isn't just 00:15:03.980 --> 00:15:07.700 about tidiness. It genuinely saves hours and 00:15:07.700 --> 00:15:10.500 massively improves consistency, reduces that 00:15:10.500 --> 00:15:14.460 mental load. Always be improving. Your projects 00:15:14.460 --> 00:15:17.799 shouldn't be static. Review them, optimize them. 00:15:18.200 --> 00:15:20.100 They're living workspaces, right? They should 00:15:20.100 --> 00:15:23.259 evolve. This whole feature. It really feels like 00:15:23.259 --> 00:15:26.620 a leap. Moving from just basic AI chats to something 00:15:26.620 --> 00:15:30.700 much more like sophisticated context -aware collaboration. 00:15:31.120 --> 00:15:33.279 The time you invest in really getting the hang 00:15:33.279 --> 00:15:35.960 of chat GPT projects today, it seems like it 00:15:35.960 --> 00:15:38.639 will genuinely pay off down the line. Maybe the 00:15:38.639 --> 00:15:40.759 best advice is just start small. Totally. Start 00:15:40.759 --> 00:15:42.820 with one project right now. Upload just a few 00:15:42.820 --> 00:15:45.620 key files. Write your first clear set of instructions. 00:15:45.820 --> 00:15:47.679 And just see the difference it makes. Exactly. 00:15:47.820 --> 00:15:50.059 Experience it. Your productivity and honestly 00:15:50.059 --> 00:15:52.759 the quality of what you create. It probably won't 00:15:52.759 --> 00:15:55.559 be the same. Yeah. This feels like the point 00:15:55.559 --> 00:15:59.480 where you shift from just talking to AI to really, 00:15:59.519 --> 00:16:02.240 truly collaborating with it. Reflective, calm. 00:16:02.580 --> 00:16:04.679 Yeah. Just consider how much more deeply the 00:16:04.679 --> 00:16:06.600 AI can understand you, understand your goals 00:16:06.600 --> 00:16:08.419 as they evolve, when you actually give it a dedicated, 00:16:08.639 --> 00:16:10.820 persistent home for your work. Two secs silence. 00:16:10.840 --> 00:16:11.919 I've got it. Have too many music.