AI Agents & Automation

StreamingLLM is an innovative open-source framework designed to address the challenges of deploying Large Language Models (LLMs) in streaming applications that require processing infinite-length inputs. It introduces the concept of "attention sinks" to efficiently manage Key and Value (KV) states, allowing LLMs to generalize to infinite sequence lengths without fine-tuning. This approach prevents the performance degradation seen in traditional window attention methods when text length exceeds cache size. StreamingLLM enables models like Llama-2, MPT, Falcon, and Pythia to perform stable and efficient language modeling with millions of tokens, offering up to a 22.2x speedup over sliding window recomputation baselines. It is particularly optimized for scenarios such as multi-round dialogues where continuous operation without extensive memory or dependency on past data is crucial.

CorMetrix offers VerixAi™, an AI-powered platform designed to transform complex medical data into clear, actionable insights for medico-legal reviews. It blends clinical expertise with advanced AI to address the unique challenges of healthcare data, which is often fragmented and filled with specialized terminology. VerixAi™ allows users to query complex healthcare data using natural language, with every answer sourced directly to the original information for full traceability. The platform supports advanced workflows, moving seamlessly from raw medical data to defensible conclusions without manual data entry or information loss. It is purpose-built for attorneys, medical experts, and LNCs, providing features like natural-language Q&A, source-linked evidence with VeriSource™, integrated imaging viewers, and secure, HIPAA-compliant collaboration. VerixAi™ is SOC 2 certified and continuously improves to meet high standards of patient privacy and data protection.

StreamingVLM is an innovative AI tool designed for real-time understanding of effectively infinite video streams. Developed by mit-han-lab, it addresses common challenges in long-video analysis by maintaining a compact KV cache and aligning training directly with streaming inference. This approach efficiently avoids the quadratic cost associated with traditional methods and mitigates the pitfalls of sliding-window techniques. The system is capable of running at up to 8 frames per second (FPS) on a single H100 GPU, offering stable and efficient video processing. It has demonstrated superior performance, winning 66.18% against GPT-4o mini on a new long-video benchmark and also enhances general Video Question Answering (VQA) capabilities without requiring task-specific fine-tuning. The project provides scripts for environment setup, inference, supervised fine-tuning (SFT), and various evaluations including OVOBench and VQA tasks.

terminal-bench is an open-source benchmark designed to evaluate the performance of AI agents, specifically Large Language Models (LLMs), in realistic terminal environments. It provides a comprehensive suite of tasks that challenge agents with complex, end-to-end scenarios, ranging from compiling code to training models and setting up servers. The tool consists of a dataset of tasks, each with an English instruction, a test script for verification, and a reference solution, along with an execution harness that connects the language model to a sandboxed terminal environment. This setup ensures reproducible and practical evaluation of system-level reasoning. It is currently in beta with approximately 100 tasks, with plans for significant expansion, and welcomes community contributions for new and challenging tasks.

Trae Agent is an LLM-based agent designed for general-purpose software engineering tasks, offering a transparent and modular architecture for researchers and developers. It provides a powerful command-line interface (CLI) that can interpret natural language instructions and execute intricate software engineering workflows using various tools and LLM providers. Key features include Lakeview for concise summarization of agent steps, multi-LLM support for providers like OpenAI, Anthropic, and Google Gemini, and a rich tool ecosystem for file editing, bash execution, and sequential thinking. The agent also offers an interactive mode for iterative development, detailed trajectory recording for debugging, and flexible YAML-based configuration. It is easily installed via pip and supports Docker for isolated task execution.

TheAgentCompany is an open-source benchmark designed to evaluate the performance of LLM agents on consequential, real-world tasks within a simulated software company environment. It allows for assessing how well AI agents can accelerate or autonomously perform work-related tasks by interacting with the web, writing code, running programs, and communicating. The platform offers diverse task roles, data types, and a comprehensive scoring system with multiple evaluation methods, including deterministic and LLM-based evaluators. It features simple one-command operations for environment setup and quick system resets, making it an extensible framework for adding new tasks and evaluators. The benchmark is available on GitHub and supports integration with platforms like OpenHands.

textgenrnn is a Python 3 module built on Keras/TensorFlow designed for creating character-level recurrent neural networks (char-RNNs). It enables users to easily train text-generating neural networks of any size and complexity on any text dataset. The tool incorporates modern neural network architectures, including attention-weighting and skip-embedding, to accelerate training and enhance model quality. Users can train and generate text at either the character or word level, configure RNN size, layer count, and use bidirectional RNNs. It supports training on generic input text files, including large ones, and allows for GPU-trained models to generate text on a CPU. Additionally, textgenrnn offers a powerful CuDNN implementation for faster GPU training and supports contextual labels for improved learning and results.

Table Transformer (TATR) is a deep learning model developed by Microsoft for extracting tables from unstructured documents, including PDFs and images. Based on object detection, TATR can be trained to work across various document domains, with pre-trained model weights available for the PubTables-1M dataset. The repository also provides the official code for the PubTables-1M dataset, a large-scale dataset for table detection, structure recognition, and functional analysis, and the GriTS evaluation metric for table structure recognition. Researchers and developers can use TATR to detect and recognize tables, convert them to HTML or CSV, and train custom models for specific needs.

TileRT is an open-source, tile-based runtime engineered for ultra-low-latency Large Language Model (LLM) inference. It aims to push the boundaries of LLM latency without compromising model size or quality, allowing models with hundreds of billions of parameters to achieve millisecond-level time per output token (TPOT). Unlike traditional inference systems optimized for high-throughput batch processing, TileRT prioritizes responsiveness, making it ideal for applications like high-frequency trading, interactive AI, real-time decision-making, and AI-assisted coding. It achieves this by decomposing LLM operators into fine-grained tile-level tasks and dynamically rescheduling computation, I/O, and communication across multiple devices to minimize idle time and improve hardware utilization. TileRT currently supports models like GLM-5 and DeepSeek-V3.2 and offers Multi-Token Prediction (MTP) for efficient longer output generation.

SmartProxy is a leading global residential proxy service provider, offering access to over 100 million residential IPs across 200 countries. Engineered for reliable web data collection and AI workflows, it provides blazing-fast, enterprise-grade access to a vast network of IPs. The platform supports various proxy types including Residential Proxy, Unlimited Proxy, Static Residential Proxy, Static Data Center Proxy, and Long Acting ISP Proxy, catering to diverse business needs. SmartProxy also offers Web Scraper APIs for real-time structured data extraction and customized solutions. With features like free geolocation, real residential IPs, and no hidden fees, it ensures high success rates for scraping and automation. The service is optimized for AI/LLM data pipelines, offering stable and reliable connections for AI data operations, ad verification, price monitoring, social media management, e-commerce, and market research.

Dova Health Intelligence, formerly Docbot, specializes in AI-powered solutions for gastrointestinal (GI) health. Their platform utilizes advanced AI to analyze live image data, providing actionable insights for clinicians and researchers. Key products include DovaVision UC for granular disease severity analysis in Ulcerative Colitis from colonoscopy videos, DovaVision BE for real-time neoplasia detection and biopsy targeting in Barrett's Esophagus during gastroscopy, and DovaSound for rapid Inflammatory Bowel Disease activity assessment from intestinal ultrasound imaging. These tools aim to improve diagnostic precision, efficiency, and patient outcomes in GI care, with several products currently available for research and clinical trials, and others in development as medical devices.

Electra Vehicles, Inc. offers an AI-powered battery intelligence platform designed to optimize battery performance and accelerate the transition to sustainable power. Their EVE-AI™ Brain for Batteries provides total visibility and control, helping users cut costs, boost ROI, and extend battery lifespan while minimizing risk. The platform is applicable across various industries, including BESS (Battery Energy Storage Systems), EVs, robotics, and aviation. Key offerings include real-time monitoring, predictive maintenance, adaptive controls, and performance intelligence for applications ranging from fleet management to automotive OEMs and energy infrastructure. Electra's AI-driven BMS (Battery Management System) ensures proactive safety, reliability, and extended battery life.

Canvass AI is an advanced AI platform designed to transform disparate information into actionable insights, offering significant improvements in accuracy, speed, and cost-efficiency. It features AI Knowledge Engines that are fine-tuned for specific sectors such as Oil & Gas, Government, Finance, Healthcare, and Manufacturing. The platform provides industry-specific AI Assistants like Engineering Assistant, Patient Case Assistant, and Regulatory Clause Assistant, which are tailored to understand technical symbols, annotations, and specifications. Canvass AI ensures high accuracy through Human-in-the-Loop validation, self-learning, and performance evaluation to prevent hallucinations. It offers flexible deployment options, including on-premise and cloud, and integrates seamlessly with existing enterprise ecosystems for rapid adoption and immediate value, delivering proven results like 30-50% faster turnaround times and significant cost savings.

Flashalgo is an AI-powered platform designed to support traders in developing and optimizing algorithmic trading strategies. The tool provides robust backtesting capabilities, allowing users to rigorously evaluate their trading strategies using historical market data. This feature is crucial for refining strategies before deployment in live markets. Flashalgo supports both automated and manual trading approaches, catering to a range of trading preferences. It integrates real-time market data, ensuring that users have access to current information for analysis and decision-making. The platform also includes user-friendly analytics tools to help traders interpret performance and identify areas for improvement, making it a comprehensive solution for strategy development and execution.

Kataba (كَتَبَ) currently serves as a parking page, indicating that the domain has expired. The page directs users to log into their Spaceship account to renew the domain, warning that it will otherwise be deleted and made available for public registration. It also features information about Spaceship, a platform for building digital presences, and offers guidance on choosing domain names. Additionally, it introduces Unbox™ for connecting products and services and Spacemail, an AI-powered spam filtering email technology. The site encourages users to contact support via live chat for further questions.

TinyStories Candle Wasm Magic is an AI chatbot tool available on Hugging Face that allows users to generate unique stories based on a starting prompt. This tool provides flexibility through customizable settings for story length, creativity, and repetition, enabling users to fine-tune the output to their specific needs. It serves as an excellent resource for educational purposes, creative writing, or simply experimenting with language models. The platform is designed to be accessible, making it suitable for a wide range of users interested in AI-powered storytelling. Its availability on Hugging Face further emphasizes its role in the open-source AI community, offering a free and interactive way to engage with generative AI.

Artwo is a platform designed to streamline the process of accessing and deploying humanoid robots for a variety of tasks. It acts as a central hub where users can search for available robotic units that match their specific job requirements. A key feature of Artwo is its ability to facilitate the deployment of these advanced robots through intuitive natural language instructions, removing the need for complex programming or specialized technical skills. The platform operates on a pay-as-you-go model, offering a cost-effective solution for businesses and individuals to leverage cutting-edge robotics without significant upfront investment. This approach makes advanced humanoid robotics more accessible and manageable for diverse applications.

DeepTutor is an AI learning companion designed to provide personalized learning assistance by transforming any document into an interactive experience. Users can upload textbooks, papers, and manuals to build AI-powered knowledge repositories, leveraging RAG and knowledge graph integration. The tool features smart problem-solving with a dual-loop reasoning architecture and multi-agent collaboration, delivering step-by-step solutions with precise citations. DeepTutor also generates custom quizzes based on knowledge bases or mimics real exam styles, and offers guided learning with interactive visualizations and adaptive explanations. It supports deep research through systematic topic exploration, web search, paper retrieval, and literature synthesis, alongside AI-assisted brainstorming.

ReiseGenie AI is a free AI tool designed to help users find the most affordable holiday deals, including package and luxury trips, by analyzing user preferences for destination, travel dates, and departure airports. It aims to save users hundreds of Euros and significant time compared to traditional booking methods. Users provide their travel wishes, and ReiseGenie AI works to find suitable deals, notifying them when offers are available. The platform emphasizes that deals are often time-sensitive due to their attractiveness. ReiseGenie AI operates on an affiliate marketing model, earning a commission from bookings without adding extra cost to the customer, often resulting in substantial savings for users.

UpScreen AI is an AI-powered candidate screening tool designed to automate manual screening tasks for talent acquisition professionals. It efficiently analyzes resumes to identify the best-fit candidates for open positions. The tool provides comprehensive candidate reports and AI-driven screening summaries, streamlining the initial stages of the hiring process. UpScreen AI also integrates with existing Applicant Tracking Systems (ATS), which helps save significant time and resources for HR and recruiting teams. By leveraging AI, it aims to enhance the efficiency and accuracy of candidate evaluation, allowing recruiters to focus on more strategic aspects of talent acquisition.

UPDF AI is a comprehensive AI assistant designed to enhance interaction with PDF documents. It offers powerful features such as summarizing, translating, and chatting with PDFs, as well as converting PDFs to mind maps and engaging in multi-modal chat with images. The tool is powered by advanced AI models including OpenAI's GPT-5 and DeepSeek R1, ensuring accurate and insightful responses. It supports both single and multi-file chat, full-document translation, and the ability to extract key information from complex data. Available across multiple platforms including web, Windows, Mac, iOS, and Android, UPDF AI caters to students, educators, writers, researchers, employees, and professionals seeking to streamline document analysis and information retrieval. It also offers a free plan with limited usage and paid plans for unlimited access and advanced features.

MiniCPM-V-4 5 is an AI chatbot hosted on Hugging Face Spaces, designed to facilitate interactive conversations with both image and text inputs. Users can upload pictures and ask specific questions about their content, or simply engage in text-based dialogue. The tool is capable of understanding the context of both visual and textual information, providing relevant and informed responses. This makes it a versatile solution for various applications requiring multimodal conversational AI, from simple inquiries to more complex analytical tasks involving visual data. Its availability as a Hugging Face Space makes it easily accessible for experimentation and use.

TravelX is an AI agent designed to simplify trip planning by generating customized itineraries. Users can input their travel details and preferences to receive tailored recommendations for various aspects of their journey, including destinations, accommodations, transportation, activities, and dining. This tool aims to provide comprehensive AI-powered assistance for organizing travel, making the planning process more efficient and personalized. It is available as a Hugging Face Space, offering an accessible platform for users to leverage AI for their travel needs.

EarlyBird is designed to streamline the freelancing experience on Upwork by offering a suite of intelligent features. It provides smart job filters to help freelancers quickly identify relevant opportunities and delivers instant alerts so they never miss a potential gig. A key feature is its AI-generated proposal capability, which assists users in crafting compelling proposals efficiently, saving valuable time. The tool aims to help freelancers focus on securing jobs rather than spending excessive time on searching and application processes, ultimately enabling them to work smarter and acquire more gigs on the Upwork platform.