Introduction to Human-in-the-loop<\/strong><\/strong><\/h2>\n\n\n\nThe concept of Human-in-the-loop<\/strong> (HILP) refers to human involvement in the decision-making processes of AI systems.<\/p>\n\n\n\nAlthough the HILP<\/strong> approach has existed in IT for a long time, it has become particularly valuable within AI. Although artificial intelligence systems are highly effective today, there remain scenarios where human judgment and verification are indispensable \u2013 especially when precision and the safety of outcomes are crucial.<\/p>\n\n\n\nHow exactly does the HILP work?<\/strong><\/strong><\/h2>\n\n\n\nWhen an AI system encounters a scenario not covered by its predefined set of rules, it pauses the process (specifically, the workflow loop) to await human input. This human-provided response enables the AI agent to continue the process in the direction specified by the human operator.<\/p>\n\n\n\n
LangGraph: Building AI-driven processes<\/strong><\/strong><\/h2>\n\n\n\nAI Agent applications can be developed using the company’s basic SDK libraries to create language models. One of the most popular is the OpenAI SDK<\/a>, which is used to communicate with ChatGPT.<\/p>\n\n\n\nHowever, alongside the evolution of LLM-based software, additional frameworks were also developed. These frameworks enable integration with models from multiple providers and offer many supplementary features that simplify agent creation and orchestration.<\/p>\n\n\n\n
One of the most popular frameworks currently is LangGraph<\/strong>, a library designed for building advanced decision-making workflows using LLMs (such as ChatGPT and others). It enables seamless integration of various components, including tools, external memory, and the core concept of human-model interaction within a specific graph (workflow).<\/p>\n\n\n\nWhat is the Process Graph?<\/strong><\/strong><\/h3>\n\n\n\nLangGraph allows workflow stages to be defined as a graph composed of nodes, each with a distinct purpose.<\/p>\n\n\n\n
Below are several node types we will use in our example:<\/p>\n\n\n\n
\n- Agent Node<\/strong> \u2013 responsible for communication with the LLM.<\/li>\n\n\n\n
- Tool Node<\/strong> \u2013 a tool that searches an external knowledge base.<\/li>\n\n\n\n
- Human Decision Node<\/strong> \u2013 where the user interacts with the process by accepting or rejecting the Agent’s work.<\/li>\n<\/ul>\n\n\n\n
The RAG technique<\/strong><\/strong><\/h3>\n\n\n\nTo build our augmented LLM, we need one more technique. In the example provided below, we used the RAG technique<\/strong>.<\/p>\n\n\n\nRAG (Retrieval-Augmented Generation) extends the capabilities of an LLM’s built-in knowledge base by incorporating an additional retrieval step, which grants the model access to external data sources.<\/p>\n\n\n\n
Thanks to this, the model can generate more accurate responses, as it incorporates context from supplementary data sources, such as a company’s FAQ. The AI chatbot application can use these data sources to provide precise answers to the company’s customers.<\/p>\n\n\n\n
In our code, RAG<\/strong> allows the retrieval of documents indexed in a vector database.<\/p>\n\n\n\nNote<\/strong>: The subject of vector databases is very wide and beyond the scope of the current article. Nevertheless, they are crucial for converting text into mathematical representations of data used by LLM algorithms.<\/p>\n\n\n\nHuman-in-the-loop \u2013 LangGraph implementation<\/strong><\/strong><\/h2>\n\n\n\nLangGraph is a framework providing an SDK for two programming languages: Python and JavaScript. In our example, we decide to use Python.<\/p>\n\n\n\n
As the “brain” of our system, we used ChatGPT’s LLM in version 4o-mini<\/strong>. Therefore, to run the application, you must first generate an API key<\/a> from the official OpenAI platform. You’ll also need to create a user account if you don’t already have one.<\/p>\n\n\n\nIn addition, using the OpenAI API may generate a minor costs, especially if we choose NOT to share the conversation data for LLM training purposes. You can find more information about this here<\/a>.<\/p>\n\n\n\nDevelopment environment setup<\/strong><\/strong><\/h3>\n\n\n\n
Although the HILP<\/strong> approach has existed in IT for a long time, it has become particularly valuable within AI. Although artificial intelligence systems are highly effective today, there remain scenarios where human judgment and verification are indispensable \u2013 especially when precision and the safety of outcomes are crucial.<\/p>\n\n\n\n When an AI system encounters a scenario not covered by its predefined set of rules, it pauses the process (specifically, the workflow loop) to await human input. This human-provided response enables the AI agent to continue the process in the direction specified by the human operator.<\/p>\n\n\n\n AI Agent applications can be developed using the company’s basic SDK libraries to create language models. One of the most popular is the OpenAI SDK<\/a>, which is used to communicate with ChatGPT.<\/p>\n\n\n\n However, alongside the evolution of LLM-based software, additional frameworks were also developed. These frameworks enable integration with models from multiple providers and offer many supplementary features that simplify agent creation and orchestration.<\/p>\n\n\n\n One of the most popular frameworks currently is LangGraph<\/strong>, a library designed for building advanced decision-making workflows using LLMs (such as ChatGPT and others). It enables seamless integration of various components, including tools, external memory, and the core concept of human-model interaction within a specific graph (workflow).<\/p>\n\n\n\n LangGraph allows workflow stages to be defined as a graph composed of nodes, each with a distinct purpose.<\/p>\n\n\n\n Below are several node types we will use in our example:<\/p>\n\n\n\n To build our augmented LLM, we need one more technique. In the example provided below, we used the RAG technique<\/strong>.<\/p>\n\n\n\n RAG (Retrieval-Augmented Generation) extends the capabilities of an LLM’s built-in knowledge base by incorporating an additional retrieval step, which grants the model access to external data sources.<\/p>\n\n\n\n Thanks to this, the model can generate more accurate responses, as it incorporates context from supplementary data sources, such as a company’s FAQ. The AI chatbot application can use these data sources to provide precise answers to the company’s customers.<\/p>\n\n\n\n In our code, RAG<\/strong> allows the retrieval of documents indexed in a vector database.<\/p>\n\n\n\n Note<\/strong>: The subject of vector databases is very wide and beyond the scope of the current article. Nevertheless, they are crucial for converting text into mathematical representations of data used by LLM algorithms.<\/p>\n\n\n\n LangGraph is a framework providing an SDK for two programming languages: Python and JavaScript. In our example, we decide to use Python.<\/p>\n\n\n\n As the “brain” of our system, we used ChatGPT’s LLM in version 4o-mini<\/strong>. Therefore, to run the application, you must first generate an API key<\/a> from the official OpenAI platform. You’ll also need to create a user account if you don’t already have one.<\/p>\n\n\n\n In addition, using the OpenAI API may generate a minor costs, especially if we choose NOT to share the conversation data for LLM training purposes. You can find more information about this here<\/a>.<\/p>\n\n\n\nHow exactly does the HILP work?<\/strong><\/strong><\/h2>\n\n\n\n
LangGraph: Building AI-driven processes<\/strong><\/strong><\/h2>\n\n\n\n
What is the Process Graph?<\/strong><\/strong><\/h3>\n\n\n\n
\n
The RAG technique<\/strong><\/strong><\/h3>\n\n\n\n
Human-in-the-loop \u2013 LangGraph implementation<\/strong><\/strong><\/h2>\n\n\n\n
Development environment setup<\/strong><\/strong><\/h3>\n\n\n\n
![\"the](\"https:\/\/sii.pl\/blog\/wp-content\/uploads\/2025\/08\/graph-visualisation.png\")
![\"job](\"https:\/\/sii.pl\/blog\/wp-content\/uploads\/2025\/08\/praca-EN-k-5.jpg\")
<\/a><\/figure>\n\n\n\n