Latest AI and tech news

Services at a citizens’ office can benefit a lot from AI assistance. Starting from a simple chatbot that answers questions about building permits, parking fines and childcare benefits to more advanced applications where AI assists filing an applicati...

Today, we’re happy to announce the release of a new major version: Haystack 3.0 🎉...

The Model Context Protocol (MCP) is the connective tissue of the modern AI stack. If you are building AI agents or production RAG systems, understanding how MCP works with Haystack is one of the most direct ways to make your applications composable, ...

Every new generation of Large Language Models arrives with a bigger context window - and the temptation to use it fully. If the model can read a million tokens, why not feed it everything? In practice, more context doesn’t reliably mean better answer...

Embeddings are the backbone of modern AI applications, from semantic search and recommendation systems to Retrieval-Augmented Generation (RAG). However, most embedding models operate in a single modality, typically focusing only on textual data....

We’re making some naming and visual updates at deepset to better reflect the role Haystack already plays as a framework, a community, and the foundation of our enterprise platform....

When you think of Lufthansa, you might picture planes, airports, or global travel, but Lufthansa Industry Solutions (LHIND) is making an impact in a different way: as a full-service IT company delivering digital solutions for clients both inside and ...

When a leading company like TELUS Agriculture & Consumer Goods (TAC), with a strong presence in agriculture and consumer goods, turns to AI to streamline complex processes, it’s worth taking a closer look....

💙 Thanks to you and all of our amazing community members, the Haystack open source framework has grown into a thriving developer ecosystem, now used by thousands of organizations to power everything from simple Q&A bots to advanced enterprise agents....

Welcome to this step-by-step tutorial where we’ll build a simple Retrieval-Augmented Generation (RAG) pipeline using Haystack and OPEA. We’ll fetch the newest Hacker News posts, feed them to a lightweight LLM endpoint (OPEAGenerator), and generate c...

Haystack is an AI orchestration framework that enables developers to effortlessly build custom AI pipelines using a modular, building-block approach. However, when it’s time to take those pipelines from your development environment to production, you...

In retrieval-augmented generation (RAG) applications, the quality of the retrieved documents plays a critical role in delivering accurate and meaningful responses. But what happens when embedding similarity is not enough to get an accurate ordering o...

The latest release from DeepSeek confirms an essential truth about AI: there won’t be one model or provider to rule them all. As the field evolves, it’s evident that diverse models bring unique strengths, making a model-agnostic approach vital for de...

In this article, we’ll explore how to build an agentic Retrieval Augmented Generation (RAG) pipeline on deepset Studio, the AI application prototyping tool for developers. We’ll first build a basic RAG pipeline and then extend the pipeline with a fal...

December is here, and so is the 2nd edition of the Advent of Haystack! This holiday season, we’re inviting the Haystack community to take part in our series of challenges around Generative AI use cases. From mastering the basics of Haystack to buildi...

When building applications with Language Models, new patterns are emerging to bridge the gap between the statistical nature of these models and the deterministic logic of traditional programming. Haystack, as an AI framework, supports developers by p...

We’re thrilled to announce an exciting new addition to the Haystack ecosystem: deepset Studio! After countless requests from our community for a visual editor to create AI workflows and invaluable feedback during the beta phase, we’re officially laun...

In artificial intelligence, multimodal agents are becoming increasingly popular for their ability to understand and integrate multiple types of input, such as text and images. In this article, we’ll show you how to build a multimodal agent that can i...

In our previous article, we explored how to build and deploy two key AI pipelines in a retrieval-augmented generation (RAG) application using Haystack with NVIDIA NIM:...

1️⃣ Extract Metadata from Queries to Improve Retrieval...

For most RAG applications, where we first have to retrieve the most relevant context, we end up having to split up documents first, and index those smaller splits of documents. Reasons for this range from needing to retrieve only relevant sections of...

DeepLearning.AI and Andrew Ng hold a special place in many AI/ML engineers’ careers and development. For many engineers, they’ve played a pivotal role into breaking into AI, or extending their knowledge and capabilities by teaching about core concept...

1️⃣ Extract Metadata from Queries to Improve Retrieval...

One of the main and critical components of a retrieval augmented generation (RAG) pipeline is the embedding process, which forms the foundation for efficient information retrieval by transforming raw text into machine-readable vector representations....

It’s 2024 and yet, most models today are still primarily geared towards English speaking markets. Today, deepset and Mixedbread are jointly announcing our latest contribution towards changing that landscape: A new open-source German/English embeddi...

In this article, we will show you how to use Haystack to evaluate the performance of a RAG pipeline. Note that the code in this article is meant to be illustrative and may not run as is; if you want to run the code, please refer to the python script...

When building real-world applications based on Language Models (such as RAG), evaluation plays an important role. Recently, evaluating generated answers using powerful proprietary Language Models (such as GPT-4) has become popular and correlates well...

Retrieval-augmented generation (RAG) systems combine generative AI with information retrieval for contextualized answer generation. Building reliable and performant RAG applications at scale is challenging. In this blog, we show how to use Haystack a...

Getting your LLM application into production is a huge milestone, but that’s only the beginning. It’s critical to monitor how your pipeline is performing in the real world so you can keep improving performance and cost, and proactively address any is...

1️⃣ Extract Metadata from Queries to Improve Retrieval...

Let’s talk about how we build AI applications that can interact with, even chat to SQL databases. Heads-up that this is a mini project that I tried out, and it’s objectively a simple approach to interacting with SQL with natural language. I’m certain...

LLMs work wonders on text data. Using LLMs, you can get answers to complex questions on long documents without having to read the document or even do a CTRL+F search. But what if you work with audio or video recordings?...

Retrieval-augmented generation (RAG) is one of the most common architectures today for building LLM applications. With RAG you can very quickly build a prototype that is tailored to your use case by connecting the LLM to the relevant data sources....

Today we are happy to announce the stable release of Haystack 2.0 - we’ve been working on this for a while, and some of you have already been testing the beta since its first release in December 2023....

Hypothetical Document Embeddings (HyDE) is a technique proposed in the paper “ Precise Zero-Shot Dense Retrieval without Relevance Labels” which improves retrieval by generating “fake” hypothetical documents based on a given query, and then uses thos...

The Haystack extension family is growing so fast, it’s hard to keep up! Our latest addition is the Astra DB extension by Datastax. It’s an open source package that helps you use Astra DB as a vector database for your Haystack pipelines....

Amazon Bedrock is a fully managed service that provides high-performing foundation models from leading AI startups and Amazon through a single API. You can choose from various foundation models to find the one best suited for your use case....

Jina.ai recently upgraded and expanded the capabilities of their previous embedding model in a v2 release....

This post was originally shared through the Haystack newsletter. Subscribe now to stay in the loop on all things Haystack, including the latest updates, new features, captivating content, and upcoming exciting events! 🗞️...

Unfortunately, countless people around the world have inadequate access to healthcare. I’m lucky to have health insurance and good medical providers taking care of me. However, I still want to educate myself before walking into a doctor’s office....

In this article, we will introduce you to the new Google Vertex AI Integration for Haystack. While this integration introduces several new components to the Haystack eco-system (feel free to explore the full integration repo!), we’d like to start by ...

One of the more cumbersome aspects of creating LLM applications is model management. Especially in cases where we need to fine-tune, host, and scale the models ourselves. In this case, having options at hand can be great. Today, we’ve expanded the Ha...