Pondhouse Data Blog
LLM Document Extraction: How to use AI to get structured data from legacy documents
The more we work with AI, the more we need to extract data from documents. In a scalable manner. While conventional and machine learning based methods provide good results for most cases, especially when the data is rather technical and provides complicated layouts like multicolumn pdfs or tables, these methods often fail. Let's test whether LLMs can help us here.
Using AI directly from your database - with PostgreSQL and pgai
AI is transforming the way we create applications and interact with data. Almost every application needs some form of AI nowadays. Every data interaction leverages some sort of AI feature. In this article, we demonstrate, how one can use AI directly from a PostgreSQL database, with only needing SQL commands.
KAN: Is this the end of feedforward networks?
MLPs are the backbone of the the most powerful AI models we know, but are KAN the next revolutionary step which will replace them?
LLM Safety with Llama Guard 2
LLMs have transformed the way we interact with information. However, they also pose a safety & security risk. In this blog, we introduce Llama Guard 2, a safety tool that helps you secure your LLMs.
How to use AI to create a PowerPoint presentation?
AI and especially LLMs have been tremendously successful in generating text. In this blog post, we will show you how to use AI to create a PowerPoint presentation. And even how to tell AI to respect your PowerPoint template
What is the Google Vertex AI Agent Builder - A practical example: Google Search with Slack.
Google's new Vertex AI Agent Builder simplifies the creation and deployment of AI-powered agents, making AI development more accessible to businesses and developers. This blog describes its capabilities by creating a Slack RAG application.
BitNet: LLM Quantization at its Extreme
Microsofts 1-Bit quantization approach which boosts performance and efficency without compromising output quality
Chat with your Confluence: A Step-by-Step Guide using Airbyte, Langchain and PGVector
In this guide, we'll explore how to build an AI chatbot that interfaces with Confluence data, using Airbyte and Langchain for data integration and PGVector for storing our text embeddings. This practical walkthrough aims to provide clear, step-by-step instructions to enable you to set up and deploy your own LLM-powered chatbot
Advanced RAG: Improving Retrieval-Augmented Generation with Hypothetical Document Embeddings (HyDE)
HyDE is a technique used to improve the performance of RAG models by generating hypothetical document embeddings based on the query and using them to retrieve relevant documents from the knowledge base.
GenAI: Technological Masterpiece or Ecological Disaster?
Since 2022 GenAI changed how we are working and already assists us in many ways. But is GenAI the revolutionary pathway to our future of worklife or is their carbon footprint the shurtcut to our ecological apocalypse?