Curated Resources

Our World in Data provides comprehensive, evidence‑based data and analysis on child mortality worldwide, combining long‑run estimates from the UN and Gapminder. The resource features interactive charts that illustrate how factors such as healthcare, nutrition, sanitation, maternal health, and living standards influence child survival rates. It serves as a key tool for researchers, policymakers, and educators to monitor progress and identify areas needing intervention.

Google Learn – Grow bietet ein umfangreiches, kostenloses Lernportal, das Kurse zu KI, Cloud‑Computing, Programmierung, digitalem Marketing, Design und digitalem Wohlbefinden anbietet. Ziel ist es, Berufstätige und Unternehmen bei der Erweiterung ihrer digitalen Kompetenzen zu unterstützen, von Einstiegskursen bis zu vertiefenden Themen. Die Plattform richtet sich an alle, die ihre Karriere fördern, ihr Geschäft ausbauen oder neue Jobs in der digitalen Zukunft anstreben.

Multi AI Agent Systems with crewAI is a beginner short course that teaches you how to surpass the performance of a single large language model by orchestrating a team of AI agents through natural language. Using the open‑source crewAI library, you’ll learn to automate complex, multi‑step tasks such as tailoring resumes or planning events in a repeatable, scalable way. The 2‑hour, 41‑minute curriculum includes 18 video lessons, 7 code examples, and hands‑on projects that demonstrate real‑world business process automation.

The JAX Quickstart page offers a streamlined introduction to using JAX, covering installation steps, core concepts, and practical examples. It guides users through setting up JAX, running basic computations, and accessing more advanced features such as automatic differentiation and JIT compilation. The page is designed for quick onboarding, making it easy for new users to start experimenting with JAX right away.

The Farmer Was Replaced invites players to program and optimize a drone to automate a farm, turning a simple plot into a self‑sustaining operation. By harvesting resources and unlocking upgrades, you strive to become the most efficient farmer in the world while honing problem‑solving and coding skills. The Steam Community hub hosts discussions, screenshots, guides, and the latest updates, keeping the player base engaged with bug fixes and new content.

OpenAI’s blog post explains how the company evaluates its language models on real‑world tasks, outlining a structured framework that blends quantitative metrics with qualitative case studies. It details the datasets, test scenarios, and scoring rubrics used to assess practical performance across domains such as summarization, coding, and conversation. The article highlights insights gained from these tests and discusses how the findings guide iterative model improvements.

Logo Diffusion is an AI-powered logo maker that transforms simple text prompts into high‑quality, vector‑ready logos in minutes. With 45+ design styles and instant previews, users can generate up to four unique concepts per prompt without any design experience. The platform offers subscription-based pricing, a user‑friendly design hub, and a free trial for new users.

NVIDIA Edify, a multimodal visual generative AI platform, expands into 3D asset creation and image control, offering early access to 3D generation tools and enterprise‑grade custom fine‑tuning. The blog highlights partnerships with Shutterstock, Getty Images, Adobe, and other content creators, while noting that Edify is no longer available as a preview microservice as of June 2025.

The MCP Model Context Protocol offers a standardized framework for designing, deploying, and interacting with AI model contexts across distributed systems. This documentation page introduces MCP, detailing its architecture, server and client components, versioning strategy, and practical guidance for building and connecting to MCP servers and clients, as well as developer tools such as MCP Inspector.

LTX 2 is a next‑generation AI video generation engine that delivers native 4K resolution and real‑time audio‑video synchronization. It offers an open‑source framework where creators can convert text or images into polished videos in seconds, from short 6‑second clips to longer scenes. With advanced features and an intuitive interface, LTX 2 lets users transform concepts into professional videos effortlessly.

Exa AI offers a powerful search engine, Websets Search, designed to handle complex queries and provide fast results for AI and machine learning applications. The platform allows developers to connect their products to a robust web search via a single API, with customizable filtering options. Try Exa AI's API for free or explore the full suite of functionality across 5 endpoints.

"Build with Claude" is an API guide and development resource provided by Anthropic Academy. It offers comprehensive documentation, best practices, and tutorials to help developers start coding and build applications using the latest models, such as Claude Sonnet 4 and Claude Opus 4. This resource provides a one-stop-shop for learning how to harness the capabilities of these models and optimize results with prompting best practices.

The paper introduces iRAG, an incremental retrieval‑augmented generation system designed specifically for video content that tackles the high upfront processing costs and loss of multimodal nuance inherent to traditional RAG approaches. By encoding and indexing visual and textual cues incrementally, iRAG enables near real‑time retrieval and generation, reducing latency while preserving richer contextual information. Experiments on multimodal datasets show that iRAG improves both retrieval accuracy and generation quality compared to baseline RAG models.

Artificial Analysis offers a concise, data‑driven overview of AI model and API providers, highlighting key metrics such as cost, latency, and performance. The blog aggregates pricing tiers, feature sets, and speed benchmarks to help developers select the most efficient and budget‑friendly solution. Regular updates keep users informed about emerging providers and market shifts.

LangChain Academy: LangChain Academy offers self-paced, comprehensive courses that teach relevant skills and knowledge to succeed with LangChain products. These courses are designed to help learners build their expertise through interactive and engaging educational content. By taking advantage of LangChain Academy's courses, individuals can gain the skills they need to excel in a variety of fields.

OpenDesk’s latest blog post explores how the platform supports secure collaboration across the German public sector, aiming for 160,000 licenses by the end of 2025. It highlights success stories from large institutions such as the Robert Koch Institute while emphasizing the platform’s adaptability to smaller, strategically important organizations. The article offers practical insights for administrators looking to implement or expand openDesk in their environments.

This notebook showcases the SocietyOfMindAgent, an AutoGen tool that orchestrates a group chat internally while presenting a single unified agent to external users. By managing internal monologues, it creates a hierarchical agent structure, reliably extracts final answers from long conversations, and gracefully recovers when context limits are exceeded. It requires installing autogen-agentchat via pip.

Playwright MCP is a lightweight server that enables LLMs to automate browsers through Playwright by exchanging structured accessibility snapshots, eliminating the need for screenshots or vision models. It offers fast, deterministic interactions and integrates seamlessly with MCP clients like VS Code, Cursor, and Claude Desktop. The server leverages Playwright's accessibility tree for LLM‑friendly, pixel‑free web navigation.

The Wayback Machine is an online service that captures and preserves snapshots of websites, allowing users to view historical versions of web pages. It is part of the Internet Archive and provides free access to billions of archived pages. Users can browse, search, and download past versions to explore how sites have evolved over time.

The “Society of Mind” notebook demonstrates how Microsoft Autogen’s multi‑agent framework can orchestrate a group of conversational agents to perform code search, repository, user, issue, and pull‑request queries. It walks through setting up agents that collaborate to retrieve and synthesize information from GitHub, illustrating practical use of the library for knowledge‑driven conversations.

The Human-Centered Artificial Intelligence (HAI) initiative at Stanford University explores the development of AI that prioritizes human well-being, safety, and values. The HAI website features news, research updates, educational programs, and opportunities to get involved in the field. It serves as a hub for information on AI-related initiatives and partnerships across Stanford and beyond.

This tool generates custom colors based on the Tailwind CSS color palette. With its intuitive interface, users can easily create unique color combinations that align with the Tailwind CSS design system. Perfect for designers and developers working with Tailwind CSS projects.

This website, "Make love, not noise", appears to be a development-focused platform offering creative resources and code snippets related to front-end development and user interface design. The site likely provides interactive web experiences, animations, and JavaScript code samples.

Lumalabs Dream Machine lets users ideate, visualize, and create stunning AI‑generated images and videos simply by asking. Powered by the Luma Photon model, it eliminates prompt‑engineering, enabling fluid, iterative exploration and remixing of creative concepts. Available on iOS and the web, the platform offers an effortless, user‑friendly experience for turning imagination into reality.

The GDELT Project blog: The GDELT Project blog shares updates, news, and insights on the Global Data Event Language (GDELT) project, a large-scale effort to collect and analyze global event data. The blog features articles, videos, and announcements on the latest developments in natural language processing, big data, and artificial intelligence. It also showcases the project's applications and collaborations with leading organizations and media outlets.

This blog post explores Interactive Voice Response (IVR) technology – the automated phone systems many businesses use. Learn what IVR is, how it works, and discover how it can improve customer service and streamline business operations. It's a helpful resource for anyone looking to understand or implement IVR solutions.

"Crontab Guru" is an online tool that generates cron schedule expressions and monitors cron jobs with Cronitor. It provides a simple editor for crafting cron schedules and offers a free, self-hosted control panel to track and manage cron jobs.

Get clear, straightforward pricing for on-demand Instances, 1-Click Clusters, and Private Cloud with Lambda's AI Cloud Pricing. Choose from flexible commitment options, including on-demand, reserved capacity at discounted rates, or self-serve clusters. Prices start as low as $3.79 per GPU-hour, depending on the term of your commitment.

Flank AI automates routine legal workflows with autonomous AI agents that review, draft, and negotiate contracts, fill out forms, and answer FAQs. The platform integrates seamlessly into existing systems, offering built‑in security, audit trails, and enterprise‑grade control for reliable legal operations.

Hume’s Speech Prosody Model transforms spoken audio into fine‑grained emotional states by analyzing prosodic cues. It outputs a mapping across dozens of affective categories—from amusement to despair—enabling real‑time sentiment analysis and nuanced user experience design. Ideal for developers building empathetic conversational agents, customer service bots, or accessibility tools.

DuckDuckGo offers a privacy‑focused search experience that blocks trackers and encrypts data, ensuring user anonymity while browsing. Its browser emphasizes a clean interface and ad‑free results, giving users peace of mind and control over their online privacy.

Swift Pathway is a curated collection of videos, documentation, and tools designed to help developers learn and build with Swift. It offers an easy‑to‑navigate, step‑by‑step approach that covers Swift’s expressive syntax, type safety, and high performance. Backed by Apple’s ecosystem and an active open‑source community, it serves as a comprehensive starting point for both beginners and experienced programmers.

Matthew Berman’s YouTube channel showcases AI‑generated videos that explore a wide range of topics, from technology and science to creative storytelling. The content blends education, entertainment, and analysis, using cutting‑edge AI tools and visuals to make complex concepts accessible and engaging.

LangExtract is a Python library that uses Large Language Models (LLMs) to extract structured information from unstructured text documents. It enables users to identify and organize key details while ensuring the extracted data corresponds to the source text, with precise source grounding capabilities. The library provides flexible options for using various LLMs and models, including Cloud, OpenAI, and Local LLMs.

Mistral AI’s chat platform offers a web-based interface for interacting with its language models, enabling users to have natural‑language conversations and access AI‑driven services. The platform supports real‑time dialogue, context retention, and integration with other tools via APIs. It serves as a flexible, developer‑friendly environment for building and testing conversational AI applications.

The Google Developers “Introduction to Large Language Models” guide explains how LLMs predict and generate text by estimating token probabilities, and details the transformer architecture and attention mechanisms that enable them to process long sequences efficiently. It covers key applications such as summarization, question answering, and translation, and offers practical guidance for developers looking to integrate LLMs into their projects.

This Harvard Gazette article questions whether modern generative AI truly “understands,” using Hilary Putnam’s ant‑drawing thought experiment to illustrate the difference between mimicking and knowing. It explores how rapidly advancing AI models blur the line between simulating thought and genuinely processing meaning.

The Hugging Face pricing page details Pro and Team subscription plans, highlighting enhanced features such as increased storage, inference credits, priority queue access, and advanced collaboration tools. Pro users pay $9/month for elevated capabilities, while Team plans start at $20 per user/month, offering SSO, granular access control, audit logs, and dedicated support for organizational workflows.

Tailwind CSS is a utility‑first CSS framework that lets developers build custom designs directly in their markup using composable classes. The official site offers comprehensive documentation, a blog, a showcase of projects, and resources for getting started, making it a go‑to hub for both beginners and experienced developers.

ChefGPT ist ein KI‑gestütztes Koch-Tool, das dir personalisierte Rezepte generiert und gleichzeitig deine Kalorien und Makros automatisch verfolgt. Mit einem Foto deiner Mahlzeit erkennt die KI sofort die Zutaten, schätzt die Nährwerte und protokolliert alles ohne manuelle Eingabe. So kannst du deine Gesundheitsziele schneller und strukturierter erreichen.

The Medallion architecture is a data design pattern used in lakehouse environments to logically structure and incrementally improve data quality as it flows through Bronze, Silver, and Gold layers. By progressively cleansing and enriching the data at each stage, it supports robust analytics, BI, and AI workloads. Often referred to as a Multi‑Hop architecture, it enables teams to maintain clean, trustworthy data at every step of the pipeline.

LongRAG introduces a retrieval‑augmented generation framework that uses much larger retrieval units—4,000‑token blocks of Wikipedia—than traditional 100‑word snippets. By compressing the corpus from 22 million to about 600 k units, the retriever’s search space shrinks dramatically, allowing the reader to work with richer context and improving overall efficiency and accuracy.

GitHub Marketplace for Models offers developers a curated catalog of AI models, integrations, and services that can be directly incorporated into GitHub repositories and workflows. The platform provides tools for discovering, evaluating, and deploying models, with access to documentation, pricing, and usage metrics. Users can browse by language, framework, or use case, and integrate models seamlessly into GitHub Actions or other development pipelines.

An interactive Streamlit web application that serves as a hands‑on example of the STL library’s capabilities. Users can explore pre-built widgets, view dynamic visualizations, and experiment with code snippets that demonstrate how to integrate STL functions into a user‑friendly interface. The app provides a practical, browser‑based environment for testing and learning the STL’s features.