Creating an agent in nemo involves defining its role and personality, selecting a generative model (LLM), choosing an embedding model, and adjusting its creativity level. These settings are flexible and can be tested and refined with the help of Aura before launching the agent.

Training in nemo consists of uploading clear and relevant information (documents, FAQs, web pages, company profiles) that the agent will use as its knowledge base. The process includes defining what it should know, organizing materials, uploading them to the platform, setting boundaries, and testing responses. Keeping the content updated ensures an accurate and useful agent.

RAG (Retrieval-Augmented Generation) allows an AI to consult your own documents and data before responding, avoiding errors and “hallucinations.” In Nemo, RAG works only with the information you provide—manuals, knowledge bases, policies—to deliver accurate, secure, and cited answers. With support from the Aura agent, you can choose the best model for your content without technical knowledge, making the most of your internal knowledge efficiently.

Choosing the ideal language model for an AI agent can be as confusing as facing dozens of options in a supermarket aisle. Aura is the conversational assistant created to simplify that decision, guiding the choice between ChatGPT, Claude, or Perplexity based on needs such as speed, accuracy, cost, and privacy. More than just technology, it focuses on enhancing human capabilities, with Aura and Nemo as allies at every stage of building the agent.