[00:02] from every iron they are shouting and instruments appearing as quickly as they disappear.  This creates confusion.  Where to start?  And is it even necessary to start? Look at these examples.  People who are far removed from development are already creating
[00:15] making money from it.  For me, the secret of their success is that they realized in time how to combine their favorite business with artificial intelligence.  Here's what's happening in the labor market.  The initial position has decreased by almost 50% in 3
[00:29] years.  While the number of vacancies for e-specialists by 2025 has increased almost twofold.  This means that the world will soon change completely and it is name is Mikhail.  I work in automation.  After this video, you
[00:43] will understand what neural networks are capable of, how to use them correctly, and where to start.  You've probably heard the terms model and agent.  Let me explain the fundamental difference.  And the model is such an advanced consultant.  It
[00:57] accepts your requests and provides answers, essentially simply communicating, based on the huge volume of data previously loaded into it .  Now let's give the model the ability to reason before responding, plan its actions, and use
[01:10] the tool.  We will also get an agent.  It already receives a task as input and produces a use a computer and an agent, it also simply generates text, for example, I can create a folder by simply clicking the mouse.  Or I can write this command
[01:24] in the terminal.  This is how an agent works.  He uses tools to generate text commands.  How are they used? Many have already heard about wipecoding.  For example, take a look at my landing page.  I didn't write a single line of code, I integrated the
[01:38] online cash register and added some nice animations .  Previously, such a site would have taken a week and a half of programmer work and 100,000 rubles.  Music. Crow suna .  You can create songs and
[01:53] upload them to YouTube or Spotify, earning money through monetization.  By the way, the copyright to the generated music belongs to you.  Voice.  Just listen to how he talks on the phone now.
[02:13] human speech.  We can already automate sales, from transcription and call analytics to cold calling and customer support.   A creators are people who specialize in product photos,
[02:29] videos, and create advertisements, cartoons, and viral videos.  Analytics.  If the data is quantitative and can be presented in a tabular format, then it can be analyzed, various calculations can be performed, dashboards can be built, conclusions can be drawn,
[02:42] forecasts can be made, and, in general, it can help you make management research and development.  A striking example is the case of an entrepreneur who created an mRNA vaccine for his dog Rosie, who suffered from cancer.  The cancer was very
[02:56] advanced.  It's not a complete cure, but the dog's quality of life has improved significantly.  In Russia, the first patient with melanoma has already received a I listed are separate tools.  Our next level is a
[03:10] installs on your computer or server and performs tasks just like you would.  For example, my OpenCla automatically keeps track of my personal finances, processes medical tests, searches for DublGiz's actions, and writes and sends
[03:25] emails.  In general, the possibilities are vast.  Now let's move on to practice.   A thousands of products.  You need to fill in the name, characteristics and descriptions. Previously, a manager would spend several hours a day on such a task.  The entire
[03:39] several hours a day on such a task.  The entire 1,000 CQU project would take us months, and his head would simply swell from the routine.  My agent does this in a matter of hours and records the final data into the database himself, and the person simply has to check the
[03:52] result.  That is, here we retain control over the meaning of architecture, we are in the free time we can do more creative and necessary.  Remember that it is an exoskeleton,
[04:06] capabilities, but it does not develop the muscles underneath it .  He's also prone to hallucinations and formulaic decisions, so don't ask him to come up with a plan for your life, make a strategic decision, or pass judgment on your relationship.
[04:19] My principle is simple.  Delegate to him only what you are ready to atrophy in yourself.  But the most important semantic work must be done independently or entrusted to easiest way is to open four neural networks.  Be sure to visit official
[04:33] websites.  Don't use Telegram Battle. Open Quen, Gemini, Clot and Chat GPT. Take a task that you understand.  You must be able to distinguish a bad result from a good one.  And ask each of the four questions.  If you are
[04:46] an accountant, give me something to calculate in a spreadsheet.  If you're a lawyer, provide contracts the limitations and capabilities of the models and choose the one that's right knowing just one neural network, but being able to
[04:59] formulate a query correctly, evaluate the response, and choose the right tool for your needs.  What do I want you to remember from this video?  In 5 years, the world will not be divided intelligence and those who do not. Everyone will use it because it's
[05:13] as easy as dialing tech.  The world will be divided into those who know how to think and those who have forgotten how because they gave it over to the machine too early.  Don't forget how to do it, and also subscribe to the channel, watch my video about Open Cla and
[05:26] reels about how I apply it in my life.  And also grab the guide I put together especially for this video. There I already described in detail what neural networks can do, where they have limitations, what relevant services are available, how much they
[05:38] cost, and how to get started on your journey to process automation.