---
title: João Lopes presents on AI chatbots to the AppX team - AppX
description: The first study from AppX Labs: João Lopes built a knowledge-based chatbot on open-source LLMs and Node.js, comparing Llama 3.1 and Gemma 2, with real numbers on embedding models and answer quality.
url: https://appx-digital.com/blog/joao-lopes-presents-on-ai-chatbots-to-the-appx-team/
---

News March 2025

# João Lopes presents on AI chatbots to the AppX team

We believe that innovation is not just a tool, but an essential pillar for building the future of digital technology. With this vision, we created AppX Labs, a space dedicated to researching and developing emerging technologies. And to mark the beginning of this journey, we present a study conducted by **João** **Lopes** on **building generic knowledge-based chatbots using** **open-source LLMs** and **Node.js**.

## The evolution of AI chatbots

Chatbots are constantly evolving with advancements in **Artificial Intelligence and Machine Learning**. With the growth of **Large Language Models (LLMs)**, these tools now offer more precise and contextual interactions.

At AppX Labs, we explored this technology to create a knowledge-based chatbot capable of answering questions about specific information and storing conversation history, ensuring more contextual and efficient interactions.

## Implementation and challenges

**João Lopes’ research** involved testing different models, including **Meta’s Llama 3.1** and **Google’s Gemma 2**. Approaches were implemented to:

- Load and process text and PDF documents;
- Create efficient embeddings for indexing and information retrieval;
- Set up a Node.js server to manage interactions with the chatbot;
- Implement a conversation history that stores user interactions;
- Test different embedding configurations to optimize response time and accuracy.

The results were promising, with superior performance from Gemma 2, especially when we used the **nomic-embed-text embedding model**, which showed the best success rate in responses.

## What’s next?

This study is just the beginning of AppX Labs. Our team will continue to explore ways to optimize user experience and investigate new applications for intelligent chatbots, including:

- Implementing cache for frequently asked questions, speeding up responses;
- Testing to optimize infrastructure and hardware;
- Comparative analysis between different LLMs and embedding models;
- Enhancing usability and integrating advanced AI functions.

Read the full article **here**.

We’re just getting started! If you’re interested in cutting-edge technology, generative AI, and innovative solutions, **follow AppX Labs**. Soon, we’ll bring new insights and research on the future of digital development.
Share

## More from AppX.
[January 2026 ### What is AppX Core and how it helps you build scalable applications faster](/blog/what-is-appx-core-and-how-it-helps-you-build-scalable-applications-faster)[December 2025 ### Planning to launch an MVP in 2026? Here’s what you need to know.](/blog/planning-to-launch-an-mvp-in-2026-heres-what-you-need-to-know)[November 2025 ### AppX at the JE Europe Autumn Conference 2025](/blog/appx-at-the-je-europe-autumn-conference-2025)