Research & Education

Zenbase AI, also known as Eva, is an AI research agent designed to significantly improve the speed and efficiency of systematic literature reviews (SLRs) and evidence synthesis. It boasts a 97.2% screening sensitivity and 97.0% extraction accuracy, validated on over 30,000 documents. Eva functions as a team of specialist AI agents that screen, extract, and verify research, allowing human reviewers to adjudicate. The tool supports living reviews, continuously updating as new evidence becomes available, making it ideal for pharma, CROs, and researchers. It integrates human-in-the-loop processes, ensuring reviewer guidance and audit-trailed results, and can deliver initial answers within 48 hours.

Webhound Reports offers long-running autonomous AI agents designed for comprehensive web research and data extraction. Users can specify their research requirements and set a budget, which directly influences the scope and detail of the research conducted. The tool is engineered to cite every fact, ensuring verifiability and accuracy in its findings. Webhound continues its research process until the allocated budget is fully utilized, providing a thorough and budget-controlled research solution. This makes it ideal for individuals or organizations needing in-depth, verifiable information without constant manual oversight.

The Bosch Center for Artificial Intelligence (BCAI) serves as Bosch’s center of excellence, dedicated to integrating cutting-edge AI technologies into the company's products and services. Established in 2017, BCAI actively spearheads applied AI projects from initial concept to implementation, bridging the gap between fundamental research and real-world applications. Operating from various Bosch Research locations globally, including the United States, China, India, Germany, and Israel, the BCAI conducts cutting-edge research using one of the world's largest datasets. The focus is on creating safe, secure, robust, and explainable Industrial AI solutions. BCAI also fosters scientific exchange through collaborations with international partners, including universities and research centers, to expand its research network and engage with industry and academic thought leaders.

Paper Pilot is an AI-powered research productivity tool designed to streamline the academic research process. It leverages cutting-edge AI to provide instant summaries of research papers, helping users quickly grasp key ideas and explore deeper insights. The platform allows researchers to effortlessly search through millions of papers, find references, and access citations with unparalleled ease. Key features include an interactive AI assistant for asking questions, powerful research boards for organizing and comparing up to 50 papers simultaneously, and AI-powered writing tools with smart citation suggestions. Paper Pilot aims to simplify literature reviews, boost research productivity, and facilitate collaboration among researchers.

Misraj AI is a next-generation Arabic AI lab dedicated to building trust and measuring impact in AI adoption. From their research lab to operational products, they create a comprehensive system that enables governments and companies to leverage AI with confidence, depth, and speed. The platform offers general solutions with integrated AI, a Large Language Models Ecosystem (Kawn), interactive AI solutions, and specialization in ML/DL/NLP. Key offerings include the Misraj System for complete AI integration, Workforces for building context-aware digital agents, and specialized platforms like Adam for Islamic da'wah and Thka for Arab awareness in technology. Misraj AI also conducts extensive research, with over 15 research papers and 35 billion open data tokens available.

AI-TOOLKIT-Training is a free, web-based tool hosted on Hugging Face, designed for training FLUX LoRA models. Users can upload their images and optional captions, define specific training parameters, and then initiate the model training process. The resulting FLUX LoRA model can be saved either locally on the user's system or directly to Hugging Face, identified by a unique name provided by the user. This toolkit simplifies the process of fine-tuning models for specific tasks, making advanced AI model training more accessible. It is licensed under Apache-2.0, promoting open access and collaboration within the AI community.

Wonders AI is an AI-powered research workspace designed to make literature review simple and transparent. It guides users step-by-step through finding, organizing, and analyzing academic papers, offering access to over 550 million scholarly sources across 88,000+ journals. The platform includes features like AI-suggested keywords, Boolean search filters, AI summaries with citations, and research project boards. It supports collaboration and allows export to various formats like PDF, DOCX, and LaTeX, with auto-formatted bibliographies in styles such as APA, MLA, and Chicago. Wonders AI is suitable for researchers of all skill levels, from undergraduates to professionals, and aims to save users significant time in their research process.

CCNets introduces a sophisticated causal learning framework designed to integrate both supervised and generative learning paradigms. This platform fundamentally reimagines traditional machine learning methods by leveraging the power of causal graphs, offering a deeper understanding of relationships within data. It provides a suite of algorithms and practical examples to facilitate the application of causal AI in various research and educational contexts. Researchers and academics can utilize CCNets to explore complex causal relationships, develop more robust AI models, and advance their understanding of artificial intelligence. For specific implementation details or to explore potential research collaborations, users are encouraged to contact the CCNets team directly.

Prompting Sheets is an AI-powered add-on for Google Sheets designed to transform workflow and boost productivity. It moves beyond generic templates by allowing users to generate custom project plans and to-do lists tailored to their specific business needs. The tool offers features like an AI Chatbot for real-time data queries, customizable prompts with dynamic placeholders, easy formula insertion with `=PROMPT()`, and one-click sheet downloads. It also includes prompt optimization, the ability to include/exclude content, and data fetching/analysis from websites directly within Google Sheets. Prompting Sheets aims to simplify complex tasks and automate processes for marketers, designers, planners, business professionals, and educators.

Flash Diffusion is an open-source tool designed to accelerate any conditional diffusion model for few-step image generation. It provides an efficient, fast, versatile, and LoRA-compatible distillation method, achieving state-of-the-art performance in terms of FID and CLIP-Score for few-step image generation on COCO 2014 and 2017 datasets. The method is highly versatile, supporting various tasks such as text-to-image, inpainting, face-swapping, and super-resolution, and is compatible with different diffusion model backbones like UNet-based denoisers (SD1.5, SDXL) or DiT (Pixart-α). It also allows for training-free acceleration of existing LoRAs and can be integrated with Hugging Face pipelines and ComfyUI.

FLARE (Forward-Looking Active REtrieval-augmented generation) is a generic retrieval-augmented generation method designed to enhance AI models. It intelligently determines when and what information to retrieve by predicting the upcoming sentence. This prediction is then utilized as a query to retrieve relevant documents, especially when the predicted sentence contains low-confidence tokens. The tool is available on GitHub and requires an Elasticsearch index for Wikipedia dumps and can integrate with Bing Search for specific datasets like WikiASP. Users also need OpenAI API keys to run experiments, with options to use multiple keys for acceleration. Debugging mode is available to walk through the iterative retrieval and generation process.

ArTok is a research paper discovery platform designed for navigating the world of academic research in AI and machine learning. It allows users to find, read, and save papers from top AI/ML conferences and ArXiv. Key features include AI-generated summaries for quick understanding of complex research, semantic search to find papers by concept, and smart tags for identifying topics and methods. Users can also perform keyword and author searches, explore interactive paper cards, and save favorites with annotations. The platform offers different discovery modes like random, trending, and personalized feeds, and supports exporting papers in various formats including PDF and BibTeX. ArTok emphasizes simplicity and privacy, with all current features being free forever and no signup required.

AI4Debunk is a pioneering European project dedicated to empowering truth in the digital age by providing innovative AI solutions to combat disinformation. The platform aims to equip citizens with the tools needed to navigate the digital media landscape safely and make informed decisions, thereby safeguarding democratic values. It leverages artificial intelligence to decipher fact from fiction, fostering a more informed and resilient society. AI4Debunk is developing four human-centered AI-powered interfaces, including real-time analysis of multimodal content, community-driven reporting, and immersive experiences. These interfaces will be powered by a first-of-its-kind open-source debunking API, with solutions like a web plug-in, app, Disinfopedia, and AR/VR experiences. The project also has an educational dimension, creating didactic materials like comic books and games to teach young people media literacy and fake news detection.

The Vlaams AI Onderzoeksprogramma, also known as the Flanders AI Research Program, is an initiative by the Flemish government dedicated to advancing artificial intelligence research. This program facilitates collaboration among various research groups, enabling them to share expertise and collectively tackle the complex challenges of tomorrow. It aims to accelerate advancements in both medical and industrial sectors through intelligent AI solutions. The program's focus areas include AI-driven data science, AI in edge computing, multi-agent collaborative AI, and human-like AI, with practical applications spanning healthcare, Industry 4.0, and government services. By bringing together diverse research capabilities, it seeks to drive innovation and practical implementation of AI technologies within Flanders.

VibeVoice is an open-source frontier voice AI platform developed by Microsoft, featuring both Text-to-Speech (TTS) and Automatic Speech Recognition (ASR) models. A key innovation is its use of continuous speech tokenizers at an ultra-low frame rate of 7.5 Hz, which efficiently preserves audio fidelity while boosting computational efficiency for long sequences. The platform employs a next-token diffusion framework, leveraging a Large Language Model (LLM) for textual context and dialogue flow, and a diffusion head for high-fidelity acoustic details. VibeVoice-ASR can handle 60-minute long-form audio in a single pass, providing structured transcriptions with speaker identification, timestamps, and content, and supports over 50 languages. VibeVoice-Realtime-0.5B offers real-time text-to-speech with streaming text input and robust long-form speech generation.

XGen is a family of open-source large language models (LLMs) developed by Salesforce AI Research, specifically designed for long sequence modeling tasks. The models, including XGen-7B-4K-Base and XGen-7B-8K-Base, support sequence lengths of 4K and 8K respectively. An instruction-finetuned version, XGen-7B-8k-Inst, is also available for research purposes. These models are released for research and development in natural language processing, particularly for academic studies. Users can install the necessary tokenization package (OpenAI Tiktoken) via pip and integrate the models using the HuggingFace Hub, making them accessible for auto-regressive sampling in Python environments.

FutureSearch enables users to deploy teams of AI agents for spreadsheet-scale research and forecasting. It offers specialized agents for deep research across datasets, predicting outcomes, ranking data, classifying entities, and matching rows across datasets. The platform supports operations like ranking, classifying, researching, merging, deduplicating, and forecasting. Trusted by major companies, FutureSearch integrates with popular AI platforms like Claude.ai, Claude Code, Cursor, ChatGPT, Codex, and Gemini, providing flexible access via its Python SDK or direct integrations. It aims to make large-scale data analysis and research efficient and accessible.

Machine Learning at Scale offers valuable resources for machine learning engineers aiming to enhance their skills and career. The platform provides weekly insights covering ML system design, real-world case studies, and practical career advice. It delves into topics such as LLM inference at scale, recommendation systems, and search and ranking systems. Founded by a Google ML engineer, the content is geared towards helping professionals become x10 Machine Learning Engineers by focusing on practical application and deep dives into complex ML systems. It also offers guidance on navigating ML job markets and preparing for interviews in big tech.

SpineDAO is a DeSci (Decentralized Science) project focused on accelerating spine health from diagnosis to recovery. It leverages AI tools, large language models (LLMs), and medtech solutions to transform back pain care. The platform aims to create a modular Spine AI Suite, including models like Vertebra-1W for wellness support and Filter for referral intelligence. SpineDAO also develops software spin-offs like Vector for clinical appointment planning and Lamina for patient information and daily spine health. The mission is to provide quicker, more efficient, and personalized spine care to over 600 million individuals worldwide through decentralized and AI-powered solutions, fostering a future where data-informed care empowers both patients and practitioners.

2Ai IPCA is a research institute dedicated to the advancement of AI technologies, bringing together experts from diverse fields such as computer vision, machine learning, and natural language processing. The institute's primary goal is to create innovative AI solutions and services applicable across various sectors, including health, industry, environment, and security. As part of the IPCA (Instituto Politécnico do Cávado e do Ave), it operates within a public higher education institution, fostering academic research and development in AI.

Deep Search AI, branded as Deal Empire Official, provides a suite of AI agents designed to automate various tasks across different domains. The platform offers browser automation for online workflows and data collection, an AI image generator for instant visual creation, and an AI research assistant to summarize and extract key insights. Additionally, it includes AI human-like text-to-speech conversion and an AI task scheduler for effortless automation of repetitive tasks. The tool aims to boost productivity and reduce costs by allowing AI to handle labor-intensive processes, making it suitable for individuals and businesses looking to streamline operations and enhance efficiency.

AfterQuery operates as an applied research lab dedicated to advancing foundation model development by curating specialized data solutions. The company addresses the challenge of suboptimal data solutions in AI research by transforming expert knowledge and real-world decision-making into structured training data. AfterQuery's methodology involves capturing how experts think, including their reasoning, decisions, tradeoffs, and context, which is then used to build datasets. Their data offerings include Supervised Fine-Tuning (SFT) with prompt-response pairs and chain-of-thought reasoning, Reinforcement Learning with expert-designed prompts and grading frameworks, Agent Environments for training and evaluating agents in real workflows, and Computer Use Trajectories demonstrating human interactions with software. This approach aims to improve model performance beyond outputs, focusing on enabling models to learn from expert reasoning.

SOM AI is an AI-powered research assistant specifically developed to support students and researchers in the demanding process of thesis writing. The tool aims to streamline various stages of academic research, from initial brainstorming sessions to refining written content. Key functionalities include assisting with idea generation, providing paraphrasing capabilities to help avoid plagiarism and improve textual flow, and offering general support to enhance efficiency throughout the entire writing journey. SOM AI is designed to help users complete their theses more effectively and is noted to provide support in Indonesian, catering to a specific linguistic audience.

Arxiv-mcp-server is an open-source Model Context Protocol (MCP) server designed to bridge AI assistants with the arXiv research repository. It allows AI models to programmatically search for papers using filters like date ranges and categories, download their content, and read them in markdown format. The server also supports local storage of papers for faster access and offers a suite of research prompts for in-depth paper analysis, including summarization, comparison, and literature reviews. It can be installed via Smithery, Claude Desktop, or manually, and supports both stdio and HTTP transport for flexible deployment. The tool emphasizes security, providing mitigations against prompt injection risks from untrusted paper content.