0:00 Music 0:07 LLM model, then you talk to him, then he will give you an output based on what you send to the LLM. 0:17 And for Alama here is able to run or to send your messages or to send your content to LLM locally. 0:26 This way you can get a free LLM. 0:31 And there is a Gemini node, there is a ChargPT node, there is also Anthropic reading node, there is also Grok one, there is Olama, DeepSeek, and you can use also the node of OpenRouter. 0:46 The OpenRouter is a service allow you to use multiple 0:51 LLM available. 0:53 Instead of subscribe in ChargPT and subscription in Gemini and other subscription to other model, you can just use or pay once at OpenRouter website. 1:05 Then you can get only one API key. 1:09 Then after that, you can select any model from what they have or what they provide. 1:17 There is a node from simple memory. 1:20 The memory can work with AI agent node. 1:23 And the AI agent node, one of the most important node in anything. 1:29 So he take an input, user input, and he has also the output. 1:35 And the most interesting here, 1:37 It can use tools and memory and the chat model. 1:42 If you link here a chat model like a charge GPT and you link here a tools like email, 1:49 and you can say to this AI agent, hey, please send me an email to that one, the charge GPT will understand what he wants or what you want, then call the tool through the AI agent. 2:06 When called the tool, he can send the email using the email provided and the content and the title. 2:13 Also from the new features, 2:16 Before was only one node, call it AI agent. 2:20 And to use another agent, you have to create another node, the same node with other workflows. 2:27 And you call them. 2:28 But now they updated and use this AI agent too. 2:34 Actually is an agent. 2:36 What's a tool for another agent, for the main agent? 2:40 And you can link it from the main agent to the AI agent, and that's one. 2:45 It can have tools here. 2:48 And when you orchestrate, you will orchestrate your workflow from the main agent. 2:56 And the main agent can execute any tool from what they have as an agent tool for him. 3:04 So this is also really powerful and create a new orchestration or a new form or new a new form of a schema inside NNN called swarm. 3:16 So there is architecture named swarm architecture. 3:21 That means you have a multiple agent. 3:23 Each agent has a multiple tools and you can use those agents as a tool for the main agent. 3:31 So you can control all the workspace or all the workflows from one agent. 3:40 And why we use multiple agents? 3:43 Because we can separate the prompt for each agent.
0:00 Music 0:07 LLM model, then you talk to him, then he will give you an output based on what you send to the LLM. 0:17 And for Alama here is able to run or to send your messages or to send your content to LLM locally. 0:26 This way you can get a free LLM. 0:31 And there is a Gemini node, there is a ChargPT node, there is also Anthropic reading node, there is also Grok one, there is Olama, DeepSeek, and you can use also the node of OpenRouter. 0:46 The OpenRouter is a service allow you to use multiple 0:51 LLM available. 0:53 Instead of subscribe in ChargPT and subscription in Gemini and other subscription to other model, you can just use or pay once at OpenRouter website. 1:05 Then you can get only one API key. 1:09 Then after that, you can select any model from what they have or what they provide. 1:17 There is a node from simple memory. 1:20 The memory can work with AI agent node. 1:23 And the AI agent node, one of the most important node in anything. 1:29 So he take an input, user input, and he has also the output. 1:35 And the most interesting here, 1:37 It can use tools and memory and the chat model. 1:42 If you link here a chat model like a charge GPT and you link here a tools like email, 1:49 and you can say to this AI agent, hey, please send me an email to that one, the charge GPT will understand what he wants or what you want, then call the tool through the AI agent. 2:06 When called the tool, he can send the email using the email provided and the content and the title. 2:13 Also from the new features, 2:16 Before was only one node, call it AI agent. 2:20 And to use another agent, you have to create another node, the same node with other workflows. 2:27 And you call them. 2:28 But now they updated and use this AI agent too. 2:34 Actually is an agent. 2:36 What's a tool for another agent, for the main agent? 2:40 And you can link it from the main agent to the AI agent, and that's one. 2:45 It can have tools here. 2:48 And when you orchestrate, you will orchestrate your workflow from the main agent. 2:56 And the main agent can execute any tool from what they have as an agent tool for him. 3:04 So this is also really powerful and create a new orchestration or a new form or new a new form of a schema inside NNN called swarm. 3:16 So there is architecture named swarm architecture. 3:21 That means you have a multiple agent. 3:23 Each agent has a multiple tools and you can use those agents as a tool for the main agent. 3:31 So you can control all the workspace or all the workflows from one agent. 3:40 And why we use multiple agents? 3:43 Because we can separate the prompt for each agent.