YT YouTube2Text

YouTube2Text → Transcripts → DeepSeek acaba de lanzar su Modo CREADOR: Crea tu Harness GRATIS 💣

DeepSeek acaba de lanzar su Modo CREADOR: Crea tu Harness GRATIS 💣 — Transcript

by Alejavi Rivera · 4,994 words · 731 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00They say that when we limit ourselves to
0:02using what is already created, we also
0:03accept the limits that others have
0:05decided.
0:06DeepSeek has done it again. It has just
0:08launched something unexpected in the
0:10artificial intelligence market that has
0:12shaken the main AI platforms. Less than
0:14a week ago, it released a mode called
0:16Creator. I’ve been testing it thoroughly
0:18, and it literally allows you to create
0:20your own AI. Not just customize, but
0:22create your own bespoke one.
0:25To verify this, we will put it to the
0:27test with seven use cases that will
0:29allow you to do things like create
0:32custom spaces, add any function we need
0:34, or even software that solves real
0:36problems, among other use cases. The
0:39best part is that, as it couldn't be any
0:41other way with DeepSeek, we will be able
0:43to use everything in a totally free and
0:46unlimited way. Sounds good? Well, let’s
0:48take a look at this.
0:53Just a few days ago, DeepSeek released
0:55its own desktop application, and it
0:57already caused quite a stir, since
0:59unlike other applications like ChatGPT
1:01or Claude, DeepSeek's was completely
1:03unlocked.
1:04We could add any model, even a free one
1:07, to use it as we pleased.
1:09But now, not content with this, they
1:11have released their Creator mode, which
1:13literally allows us to add any
1:15functionality, however advanced it may
1:17be, that doesn't exist or is exclusive
1:19to paid plans of other companies.
1:22Simply by asking the AI, it will
1:24integrate, adapt, and supercharge our
1:27own application.
1:29To do this, all we will need is their
1:31application, which is this tool right
1:33here where we have the DeepSeek harness
1:35, which they recently made easy to
1:37install on both Mac and Windows devices
1:40.
1:41Once installed, we will find an
1:42interface like this one here. First of
1:45all, keep in mind that in the latest
1:46updates they are adding extra steps for
1:48you to sign in through DeepSeek, and
1:50from there they will ask you to add some
1:52credits to use it. This is not actually
1:54necessary, as we will see throughout the
1:57video, we will be able to connect any
1:59free AI model and not depend on any kind
2:01of cost.
2:03Knowing this, and as we will delve into
2:05later, keep in mind that the new thing
2:07we will be seeing here is directly
2:09changing the mode, which by default was
2:12in standard, and now we have this new
2:14one called Creator Mode, and by
2:16selecting it, it will truly allow us to
2:18do whatever we want with this platform
2:20so that it perfectly fits all our needs
2:23.
2:23And to put this into practice, let’s
2:25move on to the second block.
2:30In this second part, we are going to
2:31dive straight into practical use cases
2:33that are genuinely useful for our
2:35day-to-day lives. We will start with
2:37some simpler, useful cases and move on
2:39to others that are much more surprising
2:41and sophisticated. Starting with the
2:43first one, one of the things I would
2:45have liked to see in an AI application
2:47is being able to see in large print how
2:49long the AI has been working, and also,
2:52once it finishes, even if we close the
2:54session, to be able to see how long the
2:56task took and check its history.
2:59That is also something some tools show
3:01you at the beginning and then even
3:03delete that information, but now, with
3:04this new Deep Nostalgia feature, we will
3:06be able to generate all these types of
3:08functions.
3:11To do this, all we would have to do is
3:13click on More, select create a plugin,
3:15and upon doing so, notice that it has
3:17directly switched us to creator mode and
3:19added this short prompt at the beginning
3:21.
3:22Now, all we have to do is ask for the
3:24change we want it to make. For that,
3:26from the prompt section, I am going to
3:29paste this entire instruction where I am
3:31asking it to create a celebration when
3:33it finishes tasks, which should also
3:35include confetti, but most importantly,
3:37we are going to ask for a floating
3:39widget that displays a giant timer with
3:42minutes and seconds so we can see all of
3:44this very easily. Keep in mind that for
3:46this entire instruction, along with all
3:49the use cases I have been preparing, I
3:51will leave you a form below for this
3:53resource that I am giving you for free,
3:55with all the prompts you will see
3:57throughout this video so you can
3:59actually reuse them.
4:01That said, I would just need to click
4:03send, but before doing so, one of the
4:05things I want you to keep in mind is
4:07that, although DeepSeek now asks you to
4:09add certain credits, notice that if you
4:11click on the more section, then go to
4:13settings, and then go to the models
4:15section, from there you will be able to
4:17include any AI provider, which would
4:19also allow you to use it for free by
4:21connecting any of its models.
4:24From here, keep in mind that I have the
4:26official DeepSeek API and I have also
4:28created an application to connect and
4:30log in with my own ChatGPT account,
4:32something I will also be leaving below
4:34in the description so you can use your
4:36own tokens and not have to invest any
4:38money in the APIs.
4:40Keep in mind that from here, by clicking
4:42on add provider, we can always add any
4:44other one that offers free APIs, such as
4:46, for example, OpenRouter, which is this
4:48platform that unifies all artificial
4:50intelligence models.
4:53Among them, if we search with the word "
4:55free," we will find a ton of available
4:57models, so by clicking on new API key
4:59after logging in, entering an example
5:01name, a limit, and hitting create, we
5:03would only have to copy, paste from here
5:05, and upon applying, we would have it
5:07available right here.
5:09Now, if I go back to the chat section,
5:11notice how beyond the official DeepSeek
5:13provider I had and the ChatGPT models, I
5:15will now be able to find a bunch of
5:17additional models we have from
5:18OpenRouter.
5:19Furthermore, if I type the word "free"
5:21here, I will be able to find all those
5:23models that I can use for free. However
5:26, that is just one method, so down in
5:27the description, in another video where
5:29I talked in detail about how we could
5:31use DeepSeek with all the free methods,
5:33you will also be able to dive deeper.
5:35Nevertheless, in my case, for this first
5:37example, I am going to be using the GPT
5:406.1 Sol model.
5:41I am going to change the effort level to
5:43medium, and if I click send, it will
5:45start processing this, and look, about 9
5:47minutes and 8 seconds later, I would
5:49have that widget we requested right here
5:51, and this also plays a victory sound
5:53effect when it reaches the end.
5:55We also get a breakdown of all the tasks
5:58it is performing, and besides, this tool
6:00you see here doesn't just stay in this
6:02session, but is already developed for
6:04this entire platform.
6:06In fact, notice that if I click start a
6:08new chat from here, it would mark
6:10everything as zero, but if I tell it
6:12anything now from here, like, for
6:13example, to briefly tell me two ways to
6:15make money with DeepSeek, upon clicking
6:17send, that large counter would start
6:19right from here.
6:21Notice how we could see it working here
6:23, and it would have automatically
6:25finished with that response in 4 seconds
6:27.
6:29Now, additionally, if I go back again to
6:31the previous chat, I can find how all
6:33this information has been saved, and
6:35with this, we can really develop
6:37anything we can think of for our own
6:39work environment.
6:42Now, imagine we want to see the usage
6:45tracking for each of the artificial
6:47intelligence models and want to see it
6:49at a glance, since platforms like
6:51ChatGPT or Claude often hide this type
6:53of information more and more, but from
6:55here, we will be able to view it however
6:57we want.
6:59To see this, I've gone to a new chat,
7:01and now from here, I'm going to click on
7:03the plus sign again to create a plugin,
7:06and I'm going to write this whole prompt
7:08where I ask it to create a plugin so I
7:10can see the usage of each AI model.
7:13Here, I'm also adding some more
7:15information that you'll find in the PDF
7:17I'm sharing via the form, so let's click
7:19send, and it will start working; and
7:21just like that, about 20 minutes later,
7:23it tells us everything is ready, and
7:25indeed, I can see the exact application
7:28I requested right here, and if I click
7:30on it, I can find all the information in
7:32detail, like the last model we used
7:34being the 6.1 Sol, all the sessions I've
7:36been using, and the entire history.
7:39We can also see the usage of each model
7:42, such as GPT-6 Sol, DeepSeek Flash,
7:44some free ones like OpenRouter, and
7:46ultimately, regardless of the model I am
7:49using, directly from this DeepSeek
7:51harness, I will be able to see the full
7:53breakdown clearly and transparently just
7:56by having asked for it.
8:01In this third case, we will see how it
8:03can also complement us in everything
8:05related to our projects and tasks that
8:07we are working on with our artificial
8:08intelligence. To do this, I have
8:10returned to a new chat, and from here,
8:12first of all, I am going to minimize
8:14this time. Notice how it would look with
8:16this new interface, and now in the chat
8:18section, imagine I would like to have a
8:20simple global view of all the tasks I
8:22want to perform, both those pending, in
8:25progress, and those I have finished.
8:28So now, if I click again on create a new
8:30plugin and add all this instruction here
8:32in the prompt section, where I basically
8:34ask it to create an application called "
8:36my space" that can be accessed from the
8:38sidebar and consists of a kind of simple
8:40, modern, and minimalist Kanban board so
8:42we can see all the tasks, whether they
8:43are in ideas, in process, or completed.
8:46Upon clicking send, well, after about 10
8:49minutes, and I say 10 minutes because
8:50even though you might have seen this
8:52minute here, notice that I had to do a
8:54small iteration to get this sentence to
8:56appear, and really, you'd have to add
8:58the other 9 minutes it took.
9:01Now, if I click on the sidebar and go to
9:03the "my space" section, I can see this
9:06application that shows us these simple
9:08statuses for each of the tasks.
9:12We could make this collaborative with
9:14artificial intelligence so we can see
9:16what it's working on at any given moment
9:18, or in a more manual way, we could
9:20click "new card," give it a name as an
9:22example, like "prepare video," I'll add
9:24the description "VIP SIP creator mode,"
9:26I'll change it to "in progress," we'll
9:28set a date, a checklist too, and once we
9:31have all that, I'll hit save.
9:34When I do this, notice how it has
9:36created this task here, which I can also
9:39change the status of, access, and mark
9:41as completed as I finish all these types
9:44of tasks in a very simple way, always
9:47having it visible in my own harness.
9:51One of the things we can do to customize
9:52the entire experience is to literally
9:54change the whole style and
9:55personalization that this harness has.
9:58To see this, I've gone back to the
10:00prompt section and now I'm going to
10:02switch to creator mode and enter the
10:04following prompt, where I tell it that I
10:06would like the entire harness to now
10:08have a retro AI OS visual style via an
10:10installable plugin.
10:12By default, I'll ask for an operating
10:14system with a 90s pixel art style, but
10:17in addition to that, I'll also tell it
10:19to add a selector with other styles that
10:22it might want. I click send, and about
10:2410 minutes later, it would tell us it's
10:26ready. Notice how we actually have this
10:29"retro AI OS" section here, and if I
10:32click "on" from here, it would have
10:34already changed completely.
10:37Different styles appear here as well
10:39that I can switch to, such as the Amber
10:41Terminal, which would look like this
10:43visualization here, or similarly, I
10:45could change it to the classic 95 style
10:46.
10:48On top of this, I can disable some
10:50effects from here, as well as select
10:53some type of mascot. Similarly, if there
10:56were any plugin you wanted to disable,
10:58even if just for a moment, beyond being
11:00able to do it from the app itself—for
11:02example, I could have disabled a mode
11:04from here and I can also restore it—
11:06well, along with that, you’ll also be
11:08able to disable it. To do that, if we go
11:10to the plugins section and enter any of
11:12them, like this stopwatch I created in
11:15the first case, if I click here to
11:16deactivate, then this would
11:18automatically no longer be active. In
11:20the same way, if I click to enable it
11:22again, it would be available here once
11:24more. With what we have just seen,
11:26personalizing artificial intelligence is
11:28becoming easier, but what if we could go
11:30a step further and automate our work?
11:32This is where Manus comes in with its
11:34latest 2.0 version, which includes some
11:36very interesting new features. The first
11:38one is automations, which allow you to
11:40execute tasks on a recurring basis. For
11:42example, researching the latest news in
11:44our industry every Monday and preparing
11:46a report for us. With that, you start
11:48configuring it and that's it, we’d have
11:50everything ready, but we can even
11:51trigger them when something specific
11:53happens, like receiving an email from a
11:55client, analyzing their request—for
11:56example, if it's an invoice—and having
11:58all the replies prepared for us.
12:00And the best part is that it doesn’t
12:02just notify us; it can actually do the
12:04work for us. Furthermore, they now have
12:07Flex Mode, which allows us to connect
12:09our own API key from providers like Open
12:11Router and choose different models,
12:13including DeepSeek, like the one you see
12:16here. This way, we can combine our
12:18favorite AI models with the tools,
12:21planning, and execution capabilities of
12:23Manus.
12:25Also, thanks to Manus and AI for
12:26sponsoring this video, they have
12:27provided a link that I will leave for
12:28you down in the description so you can
12:30try it for yourself.
12:32And now that we know this, let’s move on
12:33to the next block.
12:38In this next case, we are going to see
12:40how we could actually apply a series of
12:43more complex or sophisticated tools,
12:45such as 3D simulators.
12:47To do that, returning to our harness,
12:49notice that if I am here, having
12:51previously enabled creator mode, and I
12:53write this prompt where I ask it to
12:55create a plugin that is primarily on the
12:58left and acts like an interactive 3D
13:00game to help us understand how
13:02generative AI models work, upon clicking
13:04send, just about 3 minutes later it
13:06would tell us that it has created this
13:08for us.
13:10To see it, I'll click here on the 3D AI
13:12lab, and notice how we have this kind of
13:15explanatory laboratory of all the steps
13:18we need to be able to see this
13:20application.
13:22Also, as I can see here, it's totally
13:24interactive, where we'll be able to see
13:26the generation, probabilities, and
13:28attention, and on the left side, both
13:30the tokens and vectors, which I can also
13:32preview however I want.
13:34From here, it's also interactive because
13:37if I click on each of these steps, we
13:40can see how everything changes.
13:43We can also find different parameters
13:46that are modified as we make changes to
13:48their configuration, and ultimately, it
13:51can show us anything we need in a much
13:53friendlier way.
13:57Furthermore, it's not just this, because
13:59even if we don't understand the
14:00underlying technologies to develop a
14:02slightly more complex application, we
14:04could create our own tools based on them
14:07, so we don't depend on technical jargon
14:09or specific concepts to create more
14:11complex applications.
14:14To see this, from a new chat, I'll tell
14:16it to create an MCP connection with
14:19DeepWiki, which will allow it to see all
14:21public repositories and extract all
14:24kinds of information and technology so
14:26it can develop any type of tool without
14:29relying on a bunch of additional steps
14:32or information we'd have to provide.
14:35For that, I'm going to install it.
14:37We can see it enabled here, and although
14:40it's not something visual, if I start a
14:42new session here now, and if I tell it
14:44that I want it to use exclusively the
14:47MCP DeepWiki tool we enabled to
14:49investigate the DeepSeek repository, and
14:51by asking it to develop the stopwatch
14:54application again, upon clicking send, a
14:56little later we could see how it really
14:59relied on the DeepWiki, extracted a lot
15:01of information, frameworks, and ways to
15:04do this,
15:06and ultimately, it would be giving us a
15:08lot of alternatives to do it with all
15:10these types of repositories that could
15:13be leveraged to extract all this
15:14technology.
15:19In this next block, we are going to
15:22raise the level a bit more to see the
15:24preview and interaction capabilities
15:26with 3D models.
15:28For this, I'm now going to ask it to
15:30build a plugin called Escape AI, which
15:32is an interactive 3D game within the
15:35DeepSeek Harness interface itself, to
15:37test how much we could interact with
15:39these applications we generate.
15:43Also, I'm specifying that it should let
15:45me move the mouse and select some keys
15:47to be able to move around freely, and
15:49with all this, if I click send now,
15:51about 14 and a half minutes later we
15:53will have this Escape AI game here, and
15:55if I click it, look at how this appears
15:57.
15:58I'm going to enlarge it by minimizing
16:00the sidebar, and if I click on enter
16:02laboratory, we can see how I really can
16:04move freely with the mouse.
16:06Notice also the details of the lights we
16:08have above, and also, if I click on the
16:10keys, I can also move around this map.
16:13If I grab the key with the E key, I'd
16:15have it here, and now it tells me I have
16:17to find this door, click with E to open
16:19it, and from here we can see how it
16:21would really be fully functional.
16:24In this case, it would be very simple
16:26because I think this would be the end,
16:27and indeed it would tell us that we
16:29escaped in these 27 seconds.
16:31And look, with this we are no longer
16:33talking about creating very simple
16:35applications or for very specific use
16:37cases, but literally we will be able to
16:39make 3D simulators that are interactive
16:42, like this game, and it will be able to
16:44run smoothly. Let's put it to the test
16:46now with something a bit more complex
16:48and based on real data.
16:50For this, I am going to ask it to create
16:52a kind of real interactive world from
16:54which I can see in real time what is
16:56happening in each of the locations with
16:58a 3D Earth model.
17:01Something similar to the platform we
17:03were seeing in this video here, but
17:04let's see what we could do with a first
17:06prompt.
17:09To do this, I'm going to go to the
17:10document so you can see how it works,
17:12I'm going to select this last prompt,
17:14which is the one we are going to be
17:16using, so I'm going to click view and
17:17the prompt we could copy would land
17:19directly here, and if we click send, in
17:21just about 12 minutes later it would
17:23tell us it has created it.
17:25Also keep in mind that what we are going
17:27to see now should show real data, since
17:29I told it to consult these URLs, such as
17:32all information related to earthquakes
17:34or natural events.
17:36So returning to the application and
17:38selecting this tool here, we can see
17:40this 3D world.
17:43And besides, if I click here, notice how
17:45it’s fully interactive, and if I go to
17:47any location, like Portugal for instance
17:49, notice how I can select the zoom here
17:52, and by adding a little more, notice
17:54that if I click on this area here, it
17:56would tell me that a fire is currently
17:58happening, providing all the key data in
18:01detail so we can see it right here.
18:04Notice also that what it's showing us
18:05includes storms, fires, and earthquakes
18:07, and I can filter by those categories
18:09from right here.
18:11For example, if I only want to see
18:13earthquakes, I'll remove all the other
18:16events, and with that, I can now see
18:18that, for example, we have an earthquake
18:21here, and if I click it, it shows how
18:23Puerto Rico had a 2.8 magnitude
18:25earthquake, giving us the exact
18:27coordinates and all the details that
18:29appear directly in this section; in
18:32short,
18:34we have a 100%functional application
18:36collecting real data from the source we
18:39specified without having to develop
18:41practically anything, with the ease of
18:43always having it available as a shortcut
18:45if it's useful for our workflow or daily
18:48routine.
18:54Let’s move on to the penultimate use
18:56case, and this is where things get
18:58serious, because now imagine you need
19:00some kind of software for your
19:01day-to-day life or workflow.
19:04For example, in my case, since I upload
19:06a lot of content to social media, it
19:08would be great to have a thumbnail
19:10generator that lets me see different
19:12variations easily.
19:14Well, if I go to the creative studio now
19:16and select this prompt where I basically
19:18give it specific details, like wanting
19:21it to develop a visual application to
19:23generate 16/9 YouTube thumbnails and
19:25allow me to easily customize everything
19:27—titles, fonts, backgrounds, color
19:29palettes, templates, the ability to
19:31upload images, export in any other
19:33format, and other such information—once
19:35I paste it into a new chat and click
19:37send, a little over 15 minutes later, it
19:39would tell us that everything is done.
19:43So, let’s click directly on the creative
19:45studio section, I'll enlarge it here,
19:47and I’m already seeing things that are
19:49starting to surprise me because, on one
19:51hand, I can see different templates that
19:54, if I click them, switch between a
19:55bunch of options just as you can see
19:57here.
19:59I’m going to go with this one. It also
20:00gives us different textures that I can
20:02change with just one click. In this case
20:04, I’m going to stick with this one here
20:06.
20:07I could also change the color palette,
20:09which by default—I don't know how you
20:11brought in my academy's colors, which by
20:14the way, if you don't know it, I'll
20:15leave a link below in the description in
20:18case you want to get more serious about
20:20AI training—but look at how I can change
20:22the angle, it also lets me change the
20:24speed and intensity of everything that
20:26appears here, and I also have different
20:28palettes, like a fire palette, a sunset
20:30rose, cyber blue, and in short, a bunch
20:32of other colors,
20:34and the truth is that all of this is
20:36working perfectly and fluidly. Besides
20:38this, I can always go back to whatever
20:40palette I want.
20:42From here, I can also change how I want
20:44each of these titles to look, with a lot
20:46of details and adjustments, and if I
20:48change the word, instead of the AI
20:50changing everything, I put this here and
20:52it changes everything.
20:55Look at how I would have it modified
20:56over here so that all the changes can be
20:58applied.
21:00And the thing is, we're not talking
21:02about creating simple mini-apps, but
21:04literally being able to create these
21:06types of platforms to help us with any
21:08kind of action.
21:10In this case, look at how we would have
21:12created this mini image editor
21:14specifically for thumbnails, but let's
21:16really put it to the test with something
21:18even more complex, like having it allow
21:20us to edit video.
21:22To do this, look at how, by returning to
21:24the prompt section—and a quick side note
21:26, because look at how it added a little
21:28shortcut here, in this case for the
21:30thumbnail creative tool.
21:32In fact, if I click it, look at how it
21:33takes us back here. This is something we
21:36could change, modify, or even add any
21:38other shortcut wherever we want.
21:41And now, returning to the prompt section
21:43, if I now select the prompt called "
21:44minicat," look at how we are going to
21:46ask it for a simple mini video editor
21:48that allows us to import MP4 videos,
21:50preview them, trim them, reorder them,
21:52and do those types of basic and simple
21:54edits.
21:56So I paste it here, we hit send, and
21:58after about 12 minutes it would tell us
22:00it's already finished, so I'm going to
22:02click on this tool called "minicat" and
22:05from here, look at how we already have a
22:07video as an example. I'm going to
22:09enlarge this so I can see it.
22:11I will also be able to preview this one
22:12, just as we can see right here.
22:14I will also be able to click on import
22:16video, select the other video I had
22:18imported again, and look at how I will
22:20be able to trim it. If I click on the S
22:23, the clip is made right here, I can
22:25also reorder it, and basically make all
22:27these simple changes that might not
22:29require downloading a video editor and
22:31learning how to use it, but instead
22:33going straight to the point with what I
22:36need. Here I can also see that other
22:38different settings appear, and next to
22:40them I will also have both the download
22:43button and even the export button.
22:46And these are the small advances that
22:48could even save us a lot of time when we
22:51have to research a tool for a specific
22:53video editing task we need to do,
22:55allowing us instead to develop something
22:57much simpler that matches our needs so
22:59we can move forward faster.
23:06Let's finish with the last block, and
23:08here we will be able to see if we can
23:10really have full control of the harness
23:12as if it were truly our own.
23:14To do this, in a new chat, I'm going to
23:17write this next prompt here, where I
23:19will ask it to develop a mission control
23:22to see in detail everything we have
23:24directly in this tool.
23:26I'll click send with this, and about 19
23:28minutes later, it would tell us here
23:30that we have this application. If I give
23:33it a click, look at how we will be able
23:35to see in detail everything that is
23:36happening in the tool.
23:38All the tasks would appear here: the
23:39ones we have working right now—which
23:41currently there wouldn't be any active—
23:43the ones that are pending, and the
23:44completed ones.
23:45Also, notice how we could see the
23:47details of each of these tools, and
23:49furthermore, if I select any of them,
23:51like this one here that we just made, it
23:53would take us directly to that chat.
23:56Along with this, it also shows us all
23:58the chats with errors, the ones we have
24:00inactive, and also if I start a new one
24:02, like asking it here to research three
24:04AI news items, upon clicking send, it
24:06would already be working, and notice
24:08that if I return to the control section
24:10, it shows me that we have an agent
24:12working.
24:14I will also be able to filter this from
24:15here, and that way I will have a global
24:17view of everything that is happening,
24:19both in real time and what has been done
24:21previously, or any error or inactivity
24:23that is occurring.
24:24Besides this, it occurred to me to see
24:26if they also give us the possibility to
24:28even change the name of this harness, to
24:30, for example, change it from DeepSee
24:32Harness to AleJavi Harness.
24:34To do this, from the prompt section, I'm
24:36going to ask if it can easily change the
24:38logo with the name DeepSee Harness to
24:40AleJavi Harness, as if the application
24:42were called that, so I'll click send and
24:44a little later, about 8 minutes later,
24:46it would tell me that the application is
24:48created.
24:49It is true that I don't see the logo
24:50changed, but if I go to the plugins
24:52section, keep in mind that from here you
24:54will be able to see everything we have
24:55installed.
24:57From here we will also be able to both
24:59deactivate and activate it, and although
25:01nothing happens, going back to the chat
25:02it tells me that when I restart it will
25:04probably be applied, so I'm going to do
25:06that, and indeed, after doing so, notice
25:08how the application name has changed to
25:10AleJavi.
25:12From here I also have a small
25:13application that, if I click it, I will
25:14be able to change the name; I'll put,
25:16for example, harness. Notice how the
25:18change would also be applied here. It
25:19also lets me modify the letter here, for
25:22example to a D, I can change the color,
25:24or I can even add an image.
25:27Here, for example, in this case, by
25:29selecting the logo I have for the
25:31academy, well, notice how it would have
25:33been modified from here, and now,
25:35wherever I am, we would have completely
25:37changed the name and logo of the
25:39platform as if it were truly our own.
25:43The truth is that with this we can
25:44create everything, because customizing
25:46is adapting what others have created,
25:48but creating is stopping relying on what
25:50others have decided.

About this transcript

This page contains the full transcript of DeepSeek acaba de lanzar su Modo CREADOR: Crea tu Harness GRATIS 💣 by Alejavi Rivera, generated from the public captions YouTube serves with the video. The transcript has 4,994 words across 731 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.