00:00:10 — 00:00:15 Welcome. Magan AI is an AI interface on a portable USB drive.
00:00:16 — 00:00:22 Simply plug it into any compatible Windows PC and have instant access to open-source large language models
00:00:22 — 00:00:26 and image generators, all working straight from the drive itself.
00:00:27 — 00:00:34 We'll just give this a moment to boot up while explaining exactly what MAGAN is, and more importantly, why.
00:00:34 — 00:00:37 Ultimately, MAGAN's focus is on privacy.
00:00:37 — 00:00:42 There's a lot of good reasons why working professionals, and others, might not want to share their work, their prompts,
00:00:42 — 00:00:46 or their thoughts online. Data leaks occur often.
00:00:46 — 00:00:50 And do you really want 'Big Corp' storing every prompt you've ever typed?
00:00:51 — 00:00:57 So whether you're protecting your privacy or those of your clients, a local AI like MAGAN is a pretty safe bet.
00:00:58 — 00:01:03 And with MAGAN, your work stays on the drive itself, unless you specifically move or delete it.
00:01:03 — 00:01:07 You don't need to worry about oversharing, even on shared PCs.
00:01:07 — 00:01:11 Of course, there's no monthly fees, just the cost of the product.
00:01:11 — 00:01:16 You're free to download any other open-source models and apply them in MAGAN, as well.
00:01:16 — 00:01:22 Just be sure to check out our blog posts for more information on everything MAGAN can do for you.
00:01:22 — 00:01:27 In this demonstration, I'm going to begin by showing you how to use our in-house Support Chat.
00:01:27 — 00:01:31 I'm selecting an AI model and a persona, in this case, "Magan Support."
00:01:31 — 00:01:37 In the FILES tab, I'm going to select what's already there, which is the MAGAN support documents it will reference.
00:01:37 — 00:01:45 You can upload any text documents through the files tab if your tasks require it, but you certainly don't have to do this for most tasks.
00:01:45 — 00:01:47 This is just a cool demonstration. :)
00:01:47 — 00:01:53 I'm going to go ahead and ask the first question. Typically, the first response takes a little bit longer in any new chat,
00:01:53 — 00:01:56 so I'm going to go ahead and rename it in the meantime.
00:01:56 — 00:02:03 You'll see MAGAN is now answering my query and since I'm working with documents, it's going to include any references
00:02:03 — 00:02:09 to the file or files it's accessing. This tells us exactly where we can find those answers.
00:02:10 — 00:02:16 I'm going to ask another question. You may notice these answers are coming back blazing fast right now.
00:02:16 — 00:02:23 That's because I've set my GPU layers pretty high, and my context fairly low. I just don't need a lot of context
00:02:23 — 00:02:25 size to handle this particular task.
00:02:25 — 00:02:30 It's read through the documents and it's answering the questions related to what it's specifically referencing.
00:02:31 — 00:02:34 Again, in this case, our support documents.
00:02:34 — 00:02:40 Finally, I'd like to show you how to use MAGAN to help write your own personas.
00:02:41 — 00:02:47 These are "json" files, and the reason I want to show you this is because I'm going to demonstrate how a few persona tweaks
00:02:47 — 00:02:52 can produce considerably different results in how the AI answers your prompts.
00:02:52 — 00:02:57 In this example, I asked MAGAN to create a persona by the name of 'Action Man.'
00:02:57 — 00:03:03 If you ever want to learn more about how to create personas, and images that could be assigned to them,
00:03:03 — 00:03:06 we'll have blog posts and videos on that specific topic.
00:03:06 — 00:03:10 For now, I'd like to start a new chat with a different persona.
00:03:10 — 00:03:14 I'm going to begin by deselecting the file I had in the last chat.
00:03:14 — 00:03:19 I'll switch to a different persona and start the new chat with it.
00:03:19 — 00:03:22 In this case, I'm selecting "Emily, the writer."
00:03:22 — 00:03:25 I'll rename the chat, so I can remember it.
00:03:25 — 00:03:31 I do a bit of freelance writing, but I never go quite this far in developing ideas,
00:03:31 — 00:03:37 but I do often use AI assistance in helping to set up scenes, or tweak some paragraphs,
00:03:37 — 00:03:40 or even let me know if I'm falling into traps of repetition.
00:03:40 — 00:03:48 There's any number of ways one can employ an AI to do those tasks, without compromising the integrity of originality.
00:03:48 — 00:03:55 But in this case, we're just going to keep things simple and ask Emily to help with some general ideas for a new story.
00:03:58 — 00:04:01 And there you have it: the first answer.
00:04:01 — 00:04:06 I can already tell it's not exactly what I want, so I might just ask it again.
00:04:06 — 00:04:09 Maybe come up with something new.
00:04:13 — 00:04:16 There ya go!
00:04:16 — 00:04:24 That's pretty much it. It's already spitting out another idea, and doing exactly what I asked it to do.
00:04:24 — 00:04:31 I'm going to refresh this chat list, so you can see the name that we assigned to it.
00:04:31 — 00:04:39 At this point, I could either export this chat to a pdf file, or continue chatting with it for a while.
00:04:39 — 00:04:43 You pretty much have full control.
00:04:43 — 00:04:46 Thank you very much for watching!
