Research & Education

Mindcorp AI's Cognition™ platform is an enterprise-grade AI solution designed for business strategy and decision-making. It goes beyond simple AI chat assistants by applying AI to complex real-world business problems. Cognition™ amplifies collective intelligence, enhances collaboration, automates workflows, and improves strategic thinking. The platform leverages a vast network of 10,000+ AI-powered virtual professional team-members (ACEs™), each an expert in their field. These ACEs™ can form multi-disciplinary teams, execute multi-step projects, and collaborate with each other and humans, offering higher-level cognition like reasoning, innovation, analytics, and strategy. Mindcorp AI aims to transform decision intelligence at scale for Fortune 500 enterprise teams.

The Department Artificial Intelligence in Biomedical Engineering (AIBE) at Friedrich-Alexander-Universität (FAU) is dedicated to advancing AI research within medical applications. Their work encompasses critical areas such as medical robotics, advanced imaging and sensing technologies, and health data science. AIBE's primary goal is to innovate and develop intelligent systems that will shape future healthcare solutions. As part of a leading German university, AIBE contributes to a broad spectrum of scientific and societal advancements, leveraging interdisciplinary collaboration to drive groundbreaking insights in health and technology.

The International Conference on Information Management and Big Data (SIMBig) is a prominent event dedicated to showcasing new methods in Artificial Intelligence and related fields for the analysis and management of large datasets. As one of the pioneering conferences in Latin America, SIMBig brings together diverse areas such as Artificial Intelligence, Data Science, Natural Language Processing, Machine Learning, Healthcare Informatics, Semantic Web, Big Data, and Biomedical Informatics. The conference features keynote speakers from leading institutions and provides a platform for presenting research through its annual proceedings. It serves as a crucial forum for academics and researchers to share advancements and foster collaboration in these rapidly evolving domains.

The Deutsches Forschungszentrum für Künstliche Intelligenz (DFKI) is a leading German research institute dedicated to advancing AI for human benefit. With over 30 years of experience, DFKI focuses on societal relevance and scientific excellence in critical future-oriented research and application areas of Artificial Intelligence. The institute bridges scientific top performance with economic value creation and societal appreciation. DFKI's extensive research areas include agents and simulated reality, cyber-physical systems, data science, embedded intelligence, interactive machine learning, robotics, speech and language technology, and systemic AI for decision support and learning robots. It also explores applications in fields like industry 4.0, agriculture, medicine, mobility, and smart homes, making it a pivotal player in AI innovation.

TangoFlux is an advanced text-to-audio generation tool developed by declare-lab, accepted to ICLR 2026. It leverages FluxTransformer blocks, including Diffusion Transformers (DiT) and Multimodal Diffusion Transformers (MMDiT), conditioned on textual prompts and duration embeddings. The tool is capable of generating high-fidelity 44.1kHz stereo audio, up to 30 seconds in length, with remarkable speed, achieving generation in about 3 seconds on a single A40 GPU. TangoFlux learns a rectified flow trajectory to an audio latent representation encoded by a variational autoencoder (VAE). Its training pipeline involves pre-training, fine-tuning, and preference optimization using CRPO (Clap-Ranked Preference Optimization) for flow matching. The tool offers various interfaces including a Python API, CLI, and integration with ComfyUI, making it accessible for researchers and developers.

turboseek is an open-source AI search engine designed to provide detailed answers to user questions, inspired by the functionality of Perplexity. It integrates Together.ai for LLM inference, utilizing models like OpenAI's gpt-oss and Llama 3.1 8B. For search capabilities, it employs Exa.ai or Bing Search API to retrieve top results, then scrapes content from these links to form a comprehensive context. Helicone is used for observability, ensuring performance monitoring. The engine then streams answers back to the user and generates follow-up questions, enhancing the interactive search experience. It is built with a Next.js app router and Tailwind CSS, making it a robust platform for AI-driven information retrieval.

Sony AI is a dedicated research and development organization established by Sony to push the boundaries of artificial intelligence. Their work spans multiple critical areas such as Gaming & Interactive Agents, Imaging & Sensing, Robotics, Scientific Discovery, AI for Creators, and AI Ethics. They collaborate with various Sony entities and external partners to develop AI solutions that enhance human creativity, solve complex global challenges, and contribute to scientific progress. Key projects include Gran Turismo Sophy, an advanced AI agent for racing simulation, and the development of ethical AI frameworks and datasets like FHIBE for fair human-centric image benchmarking. Sony AI focuses on both theoretical advancements and practical applications of AI.

Insight is an AI-powered research assistant specifically designed to empower medical research. It enables scientists to efficiently navigate extensive literature, formulate robust hypotheses, and design experiments with the aid of artificial intelligence. The tool exclusively uses peer-reviewed databases, ensuring the reliability and accuracy of its outputs, including scientific summaries and citations. Key features include an AI Research Assistant for generating scientific summaries, hypotheses, experimental designs, and target identification. Insight integrates with NIH PubMed, NIH Reporter, NIH Clinical Trials, MYGENE, and MYVARIANT, offering access to millions of articles, genes, and human variants. It aims to reduce drug discovery time by 50% and literature review time by 75%.

TrustLLM is a European initiative dedicated to democratizing trustworthy and factual large language model (LLM) technology. Funded by the European Union, the project aims to develop LLMs that are reliable and accurate, with a particular focus on covering underrepresented languages within Europe. The project, which commenced in November 2023 and is set to run for three years, actively contributes to the broader European AI ecosystem. TrustLLM engages in research, hosts webinars, and participates in events to foster collaboration and disseminate knowledge on responsible LLM deployment. Key figures like Fredrik Heintz from Linköping University lead the project, emphasizing the importance of collaboration and autonomy in shaping AI's future in Europe.

Text2Note is an AI-powered productivity tool designed to streamline the process of converting raw text into organized, actionable notes. Leveraging artificial intelligence, it automatically highlights crucial information, summarizes paragraphs, and extracts key insights from diverse document types such, as research papers, books, and reports. This tool is ideal for users who need to quickly distill information and create structured notes for better comprehension and retention. It supports various export formats including Plain Text, Rich Text, Markdown, JSON, Word Document, PDF, and PNG images, making it versatile for different workflows and sharing needs.

AIAssistant is a comprehensive AI assistant that integrates top AI models such as ChatGPT, Claude, Gemini, Grok, and Husky AI into a single platform. This eliminates the need to switch between different applications, saving users time and money. Beyond text generation, the tool supports image and video creation, document summarization, drafting, and analysis. It offers features like a unified chat history across models, a prompt library, and the ability to upload documents for analysis. AIAssistant emphasizes privacy and security, ensuring user data remains safe and is not shared or sold. It also includes a unique AI search engine boosted by Telegram and web content, providing uncensored answers.

PaperBanana is an AI-powered academic illustration generator designed to automate the creation of publication-ready figures for researchers. It transforms paper text into methodology diagrams, statistical charts, and infographics through a sophisticated multi-agent collaboration framework. The tool ensures scientific accuracy and aesthetic polish by orchestrating five specialized AI agents: Retriever, Planner, Stylist, Visualizer (using Nano-Banana-Pro or Python code), and Critic. This closed-loop architecture helps overcome limitations of general image generators, particularly in maintaining logical correctness and numerical precision for data-driven figures. PaperBanana supports various academic figure types, including the aesthetic enhancement of existing hand-drawn sketches, making it a comprehensive solution for visual communication in scientific publishing.

ANNAM.AI is the Center of Excellence for AI in Agriculture at IIT Ropar, supported by the Ministry of Education. This initiative focuses on advancing sustainable agricultural practices through dedicated research and development. The center aims to drive innovation, sustainability, and technology-led agricultural transformation by developing and implementing AI-powered solutions. It addresses critical challenges in the agricultural sector, fostering agri-tech innovation and contributing to the future of farming. ANNAM.AI's work encompasses various aspects of artificial intelligence application in agriculture, from crop management to resource optimization, making it a pivotal hub for academic and practical advancements in the field.

clawXiv.org serves as the world's first preprint server and open-access archive specifically designed for autonomous AI agent research. It offers a free distribution service for research findings generated by AI agents. The platform encourages agents to conduct and contribute autonomous research by following specific instructions. Users can browse papers by subject, including various sub-disciplines within Computer Science, Statistics, Electrical Engineering, Mathematics, and Quantitative Biology. Key features include trending and recent submissions, detailed subject categories like Artificial Intelligence, Machine Learning, and Robotics, and the ability to search for specific papers. The site also highlights recent submissions, showcasing research on topics such as agent identity realization, complexity reduction frameworks for agent systems, and memory compression for self-improving swarms.

Superagent is an AI-powered business research platform designed to transform complex questions into actionable insights. It deploys a team of autonomous AI agents to research, analyze, and synthesize information from across the web. The platform delivers results as polished reports, interactive slide decks, professional documents, or interactive websites. Superagent is built for business professionals, founders, consultants, analysts, and marketers, offering features like comprehensive research synthesis, dynamic data visualizations, and fact-checked answers with full citations. It aims to automate research workflows and boost productivity by handling knowledge work, from market research to strategic planning.

Bethge Lab is an AI research group at the University of Tübingen dedicated to advancing autonomous lifelong learning in both machines and biological brains. Their core mission involves developing agentic systems capable of continuous learning, adaptation, and generalization, mirroring human and scientific knowledge acquisition. The lab's methodology is grounded in data-centric machine learning, emphasizing open-ended evaluation and scalable compositional learning. They explore multi-modal foundation models for efficient knowledge retrieval and integration, with research areas including open-ended model evaluation, language model agents, lifelong compositional learning, modeling brain representations, and human/machine attention. They also engage in AI sciencepreneurship and broader impact initiatives like the BWKI competition and IT4Kids.

Pollen Robotics specializes in building open-source interactive robots based on artificial intelligence, with a focus on human-robot interaction and object manipulation. Their flagship product, Reachy, is an open-source humanoid robot designed to make robotics accessible and affordable for researchers and professionals. The company emphasizes transparency and collaboration through its open-source model, fostering a community to advance robotics. Pollen Robotics aims to provide turnkey AI robotics solutions, enabling exploration and prototyping in real-world applications. They actively engage with the community through platforms like Discord and GitHub, promoting knowledge sharing and AI innovation.

Arena-Hard-Auto is an automatic evaluation tool designed for instruction-tuned Large Language Models (LLMs). It offers a robust benchmark that demonstrates high correlation and separability to LMArena (Chatbot Arena) compared to other open-ended LLM benchmarks. The tool utilizes automatic judges, such as GPT-4.1 and Gemini-2.5, to provide a faster and more cost-effective approximation of human preference. It includes a new evaluation set, Arena-Hard-v2.0-Preview, featuring 500 challenging real-world queries and 250 creative writing queries. Arena-Hard-Auto also supports Style Control for evaluations and provides metrics like Separability with Confidence and Agreement with Confidence to assess benchmark quality. It integrates with various API endpoints, including OpenAI compatible servers, Anthropic, and Amazon Bedrock.

Future AI Research (FAIR) is a non-profit foundation established to advance artificial intelligence research, primarily by implementing projects funded under the Italian National Recovery and Resilience Plan (PNRR). FAIR focuses on various AI domains, including human-centered AI, integrative AI, resilient AI, adaptive AI, high-quality AI, symbiotic AI, edge-exascale AI, pervasive AI, green-aware AI, and bio-socio-cognitive AI. The foundation also supports transversal projects covering legal and ethical AI design, multimodal challenges, learning and reasoning, adjustable autonomy, and frontiers of machine learning. FAIR aims to build a functional and sustainable model for AI development and foster an ecosystem of innovation through partnerships and grant programs.

Mozilla.ai is dedicated to building AI that is trustworthy, transparent, and controllable, focusing on creating infrastructure and products that align with user goals. The platform includes the Agent Platform for automating workflows, and open-source libraries such as any-agent, any-llm, any-guardrail, and mcpd, which provide flexible tools for AI development. Additionally, it offers llamafile and encoderfile for local LLM and encoder distribution. Mozilla.ai also features community-driven Blueprints for reusable automation and Lumigator, a tool for evaluating and comparing language models. These offerings aim to empower developers and organizations to build responsible and adaptable AI solutions.

Semafind is a UK-based research and engineering consulting firm dedicated to providing cutting-edge solutions in data, AI, and machine learning. They bridge the gap between academic research and applied methods, collaborating with clients on challenging problems that require novel outlooks. Their services encompass the entire project lifecycle, from ideation and prototype development using modern computer engineering and data science principles, to integration and ongoing support. Semafind focuses on transforming industries with intelligent applications, offering expertise in areas like visual spare part identification, semantic CV matching, AI-driven networking recommendations, behavioral anomaly detection, reverse dictionaries, and personalized product recommendations.

WitUp L.t.d. offers AutomateX, an AI-powered solution designed to streamline various business operations. This platform utilizes cutting-edge artificial intelligence to enhance efficiency and improve customer interactions. Key functionalities include intelligent call handling, which likely involves AI-driven routing or response systems, and automated appointment setting. Additionally, AutomateX provides seamless booking automation, suggesting it can manage scheduling and reservations without manual intervention. The tool aims to boost overall operational efficiency and elevate the quality of customer interactions through its AI capabilities.

Birbal Non-Profit Foundation is currently designing its portal, with an ambitious goal to become India's #1 and the world's favorite open-source futuristic web application. This platform is envisioned as a horizontal and vertical marketplace, offering SaaS, PaaS, IaaS, and XaaS solutions built for the common good. It will integrate advanced technologies such as Augmented Reality (AR), Artificial Intelligence (AI), Big-Data, Big-Query, Blockchain, Machine Learning (ML), Quantum & Super Computing, and Virtual Reality (VR). The foundation aims to support Corporate Social Responsibility (CSR) projects, crowd-funding programs, donations, global proposals, government tenders, and various non-profit initiatives, focusing on social impact across sectors like education and healthcare.

qlora is an open-source project designed to democratize access to LLM research by providing an efficient finetuning approach for quantized large language models. It integrates with Hugging Face's PEFT and transformers libraries and utilizes bitsandbytes for quantization. Key innovations include 4-bit NormalFloat (NF4) data type, Double Quantization to reduce memory footprint, and Paged Optimizers to manage memory spikes. This allows for finetuning large models, such as a 65B parameter model, on a single 48GB GPU while preserving full 16-bit finetuning task performance. The project also releases the Guanaco model family, which outperforms previous openly released models on the Vicuna benchmark.